GCTE -, 2t8i -

ss formations de type "brousse tirrée" des régions arides et semi-arides.

S p o n s o r e d b y :

ORSTOM I'lnstitut Fronyis & Recherche Scientifique pour le Développement en (oopérotion

CNRS-Programme Environnement

Concepfin moquette, réahotion N. Zerbib - ORSTOM Bondy

Distribution, structure, origine et devenir des brousses dites tigrées en Mauritanie sahélo-saharienne

Rossetti Ch.

2- Existing theones A geomorphological based "Tiger Pattern" generated by a former dune pattern in Sokoto Nothern Nigeria

Zonneveld 1.

c methods of study 1- Pattern at a regiorial scale

Vegetation pattern recognition using SPOT images at regional scale in Niger

Mougenot B., d'Herbès ]. M., Hamani S.

2- Functioning at local scale Patterning and pioneer processes of the herbaceous front in a tiger bush

Ehnnann M., Çalle S., Seghieri].

sses and models 1- Runoff and erosion processes

Surface crusts of the tiger bush Valentin C , Bellier Ç., Malam 1.. Rajot1.L.. Pichet].

The importance of surface water redistribution for the water budget of banded vegetation structures (Spain)

Bergkamp Ç., Cerda A., lmeson A., longejans].

The soi1 surface features of the vegetation striped patterns in Mapimi reserve. North Mexico and their influence on the hydrodynamic

]aneau 1.1.. Mauchamp A., Tarin Ç.

Impact of water harvesting variations along a climatic transect in Niger upon produciivity and paüems of tiger bush

Valentin C, d'Herbès].M.

The distribution of soluble cations within a patterned ground gilgai complex. Western New South Wales, Australia

MacDonald B.C.T, Melville M.D., White L.

Erosion processes and polyphase mosaic components in a banded grassland. and Western New South Wales. Australia

Dunkerley D., Brown K.

Soil water balance Water balance in a banded vegetation pattern: the case of the tiger bush in Western Niger

Galle S. Ehmann M., Peugeot C. -

Soil and soi1 water balance in the watenhed of the riier Corouol - a case study in the Sahel of Burkina Faso

Kappas M.

Rainfall inputs and direct recharge to the deep unsaturated zone of southern Niger

Bromleyj., Edmunds WM.. Fellman E., Brouwer]., Gaze S.R., Sudlow j., Taupin j.D.

Soil and hydrologic characteristics of a banded vegetation site in Niger and their effect on tree establishment

Zanguina 1.. Thurow I L . , juo A.S.R., Manu A.

Soil biotu Soil fauna and process regulation in patterned landscapes of serni-arid Australia

Noble ]. C.

Tenites-soil-vegetation interactions in a striped vegetation pattern (Burkina Faso)

Ouedraogo P. iepage M.

4- Populations dynamics Cornparison of spatial patterns of woody species in a spotted savanna and a tiger bush of Northern Yatenga (Burkina Faso, West Africa)

Couteron P

Runoff contribution to a sahelian two-phase mosaic system: soi1 water regime and vegetation life cycles

Seghieri 1.. Çalle S.

5- Syntheiic models of fundioriiiig Stripes, strands or stipples: modelling the influence of three banding patterns on resource capture and productivity in semi-and woodlands, Australia

Ludwig 1.. Marsden S.

'Brousses tigrées": an example of pattern formation in systems with non local interactions.

Lejeune O., Lefeue, R.

6 - Genesis of banding geneml niodeis Soil organic matter dynamic on Tiger Bush in Niger

Çuillaume K.. Abbadie L.

Using 137Cs to trace upslope migration of banded vegetation in South-west Niger.

Chappe11 A., Valentin C, Warren A., Charlton M. d'Herbès ]. M.

How do vegetated arcs move upward in Niger ? Mougenot, B., d'Herbès, ].M.. Ichaou, A.

Banded vegetation patterning in a subantartic forest of Tierra del Fuego, as an outcome of the interaction between wind and tree growth.

Puigdefabregas]., Çaliart E, Bianciotto O., Allogia M., Del Barrio Ç.

sequences for Management 1 - Productivity of patterned vegetation

Does patchiness increase vegetation productivity, biodiversity and stability ? The case of the 'brousse tigrée" in the Sahel

Hiemaux FI

2- Towards an iniproved land nianagenieni in the and and seni ind zones The evolution of soil-vegetation patterns following land abandonment and fire in Spain

Carnmeraat L. H.. lmeson A.C.

Tiger bush that lost its stripes: the effect of grazing on the spatial pattern of herbage production in a banded semi-arid woodland in Eastern Australia

Freundenberger D., Palmer R.

Vegetation arcs and litter dams: similarities and differences Eddy]., Humphreys Ç.S., Hart D.M., Mitchell PB., Fanning P C.

Relationshps between rainfall. soil-water, and herbage production in three geomorphic zones in semi-arid wooded rangelands

Çreene R.S.B., Noble 1. C.

des brousses dites tigrées

ette note n'a pas pour objectif de reproduire in extenso des observations fort longues effectuées en juin-juillet 94 en cours d'exploitation. mais de fournir une orientation générale sur ces observions qui avec d'autres observations portant sur une trentaine d'années doivent

illustrer certains aspects quelque peu controversés sur l'évolution de la végétation en Afrique du NW suite aux grandes sécheresses.

*:* VUE GÉNÉRALI S U R LA QUESTION DU PCllNT DE V l l E DE LA PHYTOECOLOGIE

J'ai eu connaissance de l'existence de cette structure remarquable de la végétation par des collègues australiens en 1957 lors d'exploration dans la région d'Alice Springs en Territoire du Nord. L'équipe du CSIRO (Land Research Division) effectuait alors des relevés d'inventaires globaux par la méthode combinant une orientation morphologique combinée à des relevés de végétation, à partir de couvertures aériennes à l'échelle du 150 000ème. C'est sur la photographie aérienne qu'a été, sauf mieux informé, découverte la structure particulière en bandes de physionomies mixtes. essentiellement arbustives en Australie (Acacia aneura. sauf erreur), puis mixtes en Afrique de l'Est. et enfin, par la cartographie au 1 :200 000ème de I'IGN, en Afrique du NW.

D'un double point de vue de la phytogéographie, et à une autre échelle de travail. de la phytoécologie, cette disposition de la végétation offre un exemple fort intéressant comme indicateur du milieu. Sur le plan climatique certains types de climat favorisent-ils cette disposition? Sur

* Ingénieur agronome EPF PART I

R BUSH AND RELATED STRUCTURES

un plan écologique, quelles conditions de station permettent-elles cette constitution. Enfin se pose la question de savoir comment de telles dispositions peuvent-elles apparaître au moment de leur constitution, quelle est leur stabilité et quels facteurs peuvent modifier cette constitution. Accessoirement la phytoécologie s'intéressera à la composition botanique de ces végétations. mais ce n'est pas là. à mon avis, un problème crucial, car en l'occurrence c'est la structure qui est cause d'interrogations. non la floristique, trop fréquemment exhaustive d'études phytoécologiques.

*:* LA SITUATION EN MAURITANIE D U CENTRE NORD ET DU NE

Notre préoccupation en 1959/61. il y a plus de trente ans, n'était nullement d'étudier en soi les BT. La prospection écologique engagée par la FA0 avait pour objet de rechercher en zone saharienne et sahélo saharienne les structures de végétation contrastées que l'on supposait alors favorables à la localisation pour la reproduction des essaims du criquet pèlerin (phase grégaire). Depuis la définition de ces zones par les travaux de POPOV s'est considérablement affinée. et le travail effectué sur les BT n'a. d'un point de vue acridien, qu'un intérêt tout à fait secondaire..Tel n'était point alors mon avis lors de la prospection effectuée avec P. AUDRY, pédologue de I'ORSTOM détaché au projet FA0 et avec qui un début d'investigation d'un échantillon de BT fut réalisé sur le plateau de Néma en Mauritanie de l'est.

Avant d'examiner la structure caractéristique, car plus ou moins linéaire. en bandes, ou en "arcs" selon la terminologie anglo-saxonne, tâchons, tout au moins pour l'Afrique de NW de visualiser les zones où se produit ce phénomène.

C'est vraisemblablement CLOS ARCEDUC, ingénieur géographe de I'ICN qui l'a, le premier, signalé en zone sahélienne a tenté de lui donner une interprétation. Dans les actes de la Conférence de Toulouse de 1964 (op. cit 1968) il révèle ce que l'accès libre à la photothèque (et du temps disponible !) lui permettaient de déceler : des images caractéristiques en dessein de pelage de tigre (d'où la dénomination) au Mali (entre 7 et 5" W. 15 et 16"N), dans le sud du pays Dogon toujours au Mali (3"W et O" et en écharpe entre les latitudes 14 et 16"N). enfin dans une zone d'extension considérable au Niger entre I " et 4" E et au delà ; puis en écharpe entre 12 et 16" de lat. N). II en avait fait mention précédemment dans une brève note publiée dans un bulletin de I'IFAN en 1956 que m'avait signalée Th. MONOD. mais

TlGER BUSH AND RELATED STRUC

dont je n'ai jamais disposé. C'est vraisemblablement à cette occasion que cet observateur averti de la photographie aérienne a proposé le nom de brousse tigrée dénomination qui fut reprise par les cartographes de I'lGN interprétant les photographies en termes d'images, non en termes écologiques. Je montrerai pourquoi je fais cette réserve..

Nulle mention n'est faite de la présence de BT en Mauritanie. En fait on les trouve en discontinuité sur des plateaux dans la région de Boûmdeit entre II "55'et II "30'W et 17O35'N. ceci à quelques minutes sexagésimales près. parfaitement portées sur la carte au 1 :200 000ème) comme sur les feuilles de Oualata et Néma. Sur la première avec les extensions nord extrêmes dans cette région 17" IO'par lambeaux entre 7" IO'et 7" W et sur celle de Néma .Le plateau de Néma présente un cas intéressant car les BT figurant sur la carte IGN et très bien repérés (des prospections extebsives l'ont confirmé se présentent en blocs disjoints entre les lastitudes i 6O28' et 17" puis légèrement à l'est de la longitude 7" ouest et 7'20' aproximativement. la distribution de ces blocs par rapport au support est démontrée dans la photographie aérienne reproduite ici ( 1 :50 000ème).

Je reviendrai en détail sur la distribution des BT sur le plateau de Néma dans le corps de cette note.

*:* ~ ~ N S I D É R A T I O N G E M E R A L E S SUR LES S T P U t T l I R E S GEOMETRIQUEMENT SIMPLES D E LA VEGETATION

L'existence de structures* linéaires pose un problème général d'interprétation qui se traduit dans la formulation d'une indication écologique de tels types de végétation.

II est quelque peu regrettable que la photographie aérienne ait souvent servi, non comme instrument de diagnostic à partir duquel une explication doit être recherchée sur le terrain. mais comme substitut à l'enquête de terrain. C'est par ce court circuit que des raisonnement spécieux reposant sur I'arte fact de l'image peuvent être faits, et c'est dans ce piège qu'est tombé CLOS ARCEDUC.

Le cas le plus général de la linéarité est la structure discontinue. en mosaïques, par rapport à la structure diffuse. Au niveau de la répartition des espèces ce phénomène a été, je crois, le mieux étudié par l'école de

*le rappelle que dans la description de la végétation. le terme de structure indique à la fois les caractéristiques de PART I distribution horizontales comme verticales de celle-ci. hidemment les images photographiques aériennes à petite ,-MW- échelle prises en considération dans les études de végétations qui ne justifient pas des prises de vues spéciales à @@%@ des échelles plus grandes ne révèlent que la structure horizontale, c2st à dire le mode de répartition en surface. lg#%$, .<;*:.:<,,...,v%i.i$

BUSH AND RELATED STRUCTURES

Montpellier (CEPE), mais pour des végétations relativement fermées (éléments + - jointifs). Dans le cas des végétations sub tropicales sèches, ce caractère jointif est plutôt rare ou alorsseulement saisonnie, comme pour les prairies éphémères. Dés que la pluviosité devient le facteur écologique principal qui se traduit par la disette en eau et que la concurrence pour I'eau devient le facteur écologique dominant la discontinuité est de règle, mais pas toujours, et de loin. la discontinuité de type linéaire.

MONOD a voulu distinguer répartitions diffuse et contractée supposant que cette dernière était caractéristique des végétations sahariennes. II n'en est rien. La galerie forestière, la lette derrière la dune de protection en Aquitaine sont des structures linéaires. mais n'ont évidemment rien de saharien. Le facteur qui commande la linéarité au Sahel et au Sahara est celui de l'approvisionnement en eau : soit par le relief, soit par le type de texture superficielle du sol. Le ruissellement provoque la discontinuité. le couvert sableux le caractère diffus de la répartition, avec nuances d'ailleurs comme le montre la colonisation d'espace sableux dans des stades initiaux de colonisation.. D'autres facteurs peuvent provoquer la discontinuité comme la présence de termitières, la formation de croûtes (la battance des sables argileux, ou bien .la concurrence pour I'eau et parfois des caractères de croissance comme les ocelles du Spinifex en Australie du Nord).. On ne peut se dispenser d'aller voir sur le terrain d'un œil critique l'image produite par la photographie aérienne dont l'avantage consiste ceopendant à amplifier considérablement les caractères de répartition et ainsi de les rendre plus visibles. CLOS ARCEDUC a mis en relation des brousses tigrées au Niger sans rapport écologique les unes avec les autres déduisant à tort qu'elles dérivaient les unes des autres alors qu'elles traduisaient, dans les exemples choisis. des phénomènes stationnels différents. Cependant ce travail n'a maintenant plus qu'un intérêt historique. et aucune portée pratique, notamment pas pour éclaircir certaines questions quant à la désertification dont j'aurai à parler en fin de note.

Pour ma part je suis arrivé à la conclusion générale que l'existence d'une géométrie linéaire. ou bien d'une distribution moins géométrique comme la formation d'une mosaïque mono ou polyspécifique. mais avec nette dominance ponctuelle d'une seule espèce, indique avec. de fortes présomptions. que la végétation qui présente ces types de structures horizontales est en passe de colonisation d'un espace précédemment nu, que cette phase de colonisation peut se stabiliser temporairement mais durer un temps très long (plusieurs décennies), ceci en raison de phénomènes de concurrence interne aux groupements pionniers qui

PART l "BE,; y++ ..yQ.;: $$&$ " ;;:$3&,i*3

TlGER BUSH AND RELATED STRU

colonisent l'espace ou bien par la perpétuation d'une instabilité des conditions stationnelles. Pour moi le phénomène intéressant est celui de la dynamique stationnelle e t non une instabilité zonale qui de surcroît ne tient pas compte d'un autre groupe de facteurs : l'ensemble des perturbations biotiques. Les BT appartiennent à ce type de végétation instable.

Ainsi l'insertion des BT dans le cadre général de la zonation nord sud de la végétation en Afrique du N W se situe à deuxniveaux : l'un bien sûr dans le cadre de la disposition du gradient pluviométirique (en quelque sorte la clef de la partition), mais, infiniment plus important, dans la ou les mélodies à travers la diversité de conditions stationnelles dans un cadre cependant défini. C'est par l'inversion du jeu de ces déterminantes que se situe l'erreur de CLOS ARCEDUC et d'ailleurs. dans une moindre mesure. de MONOD.

l'ajouterai en guise de remarque générale. qui aurait pu constituer une conclusion méthodologique. ceci. L'école australienne où j'ai reçu ma formation majeure ne comment pas l'erreur, comme les phytogéographes et l'ancienne école de la phytosociologie européenne, de regarder la végétation avec des yeux couverts de lunettes filtrantes dont le filtre est telle ou telle théorie. Elle la regarde comme un fait local qu'il convient de comprendre par l'étude des autres faits stationnels qui la conditionne. Elle fait ainsi progresser les études du particulier vers le général et non l'inverse, source de graves aberrations et de conclusions qui ne sont finalement que des vues de l'esprit. C'est pour cette raison qu'à ma connaissance c'est l'école australienne qui la première a vu juste dans la reconnaissance du facteur responsable pour la formation des BT. AUDRY et moi-même n'avons pas fait preuve d'imagination débordante en reprenant à notre compte leur hypothèse. simplement l'affinant, l'illustrant et la mesurant dans une station donnée sur le plateau de Néma dont je vais rappeler, parce qu'ils n'ont jamais été publiés, les principales conclusions.

Si l'on regarde les BT avec l'idée qu'il s'agit, non de groupements colonisateurs. mais de groupements climaciques. ou bien, exprimé en d'autres termes. comme des élément du paysage au même titre qu'un guelb. un reg ou même un sol, on passe à côté d'une réalité changeante. fonction de deux facteurs : majeurs :

J le macro climat par son induction de conditions de morphologie superficielle stationnelle (dynamique du ruissellement et remaniements éoliens, comme je le montrerai)

PART l :v,?v*> vi#$d&$$E $9 gg& *xi ;&

BUSH AND RELATED STRUCTURES

J.conditions de dynamique interne au groupement végétal lié principalement à des phénomènes de concurrence entre dominantes ponctuelles, définition que je donnerai.

+$ LA STATION D U PLATEAU DE NÉMA (ÉTAT 1961)~

Je disposais à cette époque non seulement d'une faculté de décision quant au choix de stations à étudier, mais également de moyens exceptionnels sous forme notamment d'un dispositif de photographie aérienne à très grande échelle**. de moyens aussi pour localiser les observations par point astronomique. Le résultat en furent les documents reproduits ici aujourd'hui pour la première fois publiés et devenus introuvables.

Les légendes des diverses figures, mais surtout celles-ci illustrent mieux que ne pourrait le faire un commentaire les principales conclusions :

J les bandes sont disposées à 90" environ de l'orientation majeure du sens de ruissellement sur des pentes très faibles semi horizontales et dans les ruptures de pente où il y a infiltration localisée de l'eau de ruissellement

J elles sont disposées en arcs, ce qui n'est pas expliqué et montrent une structuration horizontale caractéristique. à la fois en géométrie et en répartition d'espèces localement dominantes et exclusives de tout autre appelées dominantes ponctuelles. La séquence de dans la structure de la bande illustre la zonation climacique de la répartition de ces espèces dans le gradient général.

J la distance interbandes est supposée non refléter. comme l'estime CLOS ARCEDUC une évolution d'un système de bandes serrées à des bandes plus espacées, fonction du vieillissement des systèmes ou de leur situatiation dans la zonation climatique. mais un phénomène très localisé de gradient de pentes plus ou moins fort, ou bien des ruptures de pentes plus ou moins espacées. Ceci est une supposition logique découlant de l'observation sur une seule bande. car le prouver eût demandé un travail majeur de recherche qui n'était pas l'objet de l'exploration d'alors.

J les bandes montrent une structure verticale parfaitement illustrés par le schéma ci-joint.

PART , voir bibliographie 2qwg,, " décrit dans la note présenfie en bibliographie F$$f$?~>*;E .r..- *,<?r:" i*:,,>. ..: . <>>A" g&**&n.$$ ;%@$gg3&

TlGER BUSH AND RELATED S'[RU

*:* LA MÊME STATION À TRENTE A N S D'ÉCART

L'orchestration généralisée autour de la pseudo désertification à partir de vues de l'esprit en l'absence d'observations sérieuses de terrain m'a ce point choqué que j'ai décidé en 1988 de financer moi-même une série de trois voyages destinés à établir la VÉRITÉ (que certains désignent par le terme quelque peu ridicule de vérité au sol. le "ground truth" américain comme si de vérités il y avait de deux catégories) par rapport à l'état en général fort bien documenté, et en tout cas parfaitement localisé de stations photographiées par ballon et au sol et quantifiées. ce qui est primordial par rapport aux observations qualitatives inchiffrables et non comparables*.

II n'est pas possible ici de présenter I'ensemble des résultats de deux de ces missions (l'une ayant été un échec personnel), mais il faut en limiter la portée pour une bonne raison : ce genre de travail quantifié requiert des moyens assez considérables qui ne sont pas à la portée d'un privé. Et compte tenu de l'irresponsable gaspillage international et national de fonds opéré par l'orchestration à tonalité apocalyptique engendré par les Nations Unies dans le domaine de la pseudo désrtifiction. concept flou. voire spécieux, il n'est plus aujourd'hui possible de justifier de telles études au niveau d'administrations constipées par l'excès antérieur de dépenses ensablées dans des études sans objet précis. sans référence aux observations sérieuses passées et non archivées, etc. etc. Je n'aurais pas de mots assez crus pour stigmatiser l'irresponsabilité d'une agence des Nations Unies dans ce domaine : I'UNEP en principe chargée de coordonner l'ensemble des activités dans ce domaine, et ceci doit être dit et connu car le gaspillag de fonds devenus rares se poursuit impunément, et non seulement à I'UNEP, mais en France, par I'OSS.

Donc en man avril 1988. doté de moyens qui n'étaient autres que ma mémoire et mes pieds, je pus retourner sur le site de Néma. mais je ne pus retrouver le point précis d'observation de i 961. Cependant à ma grande surprise et en bordure du plateau, et non loin du site en question la brousse tigrée avait disparu au profit d'une colonisation diffuse par du Commiphora africana sous fonne d'arbustes vigoureux et jeunes. L'état des lieux qui avait changé de fond en comble était la couverture par une mince pellicule de sable du reg gravillonnaire antérieur. Apparemment il ne subsistait pas grand chose des bandes antérieures.

cette tendance de la phyto-écologie française à se limiter aux obsenibations qualitatives a heureusement été PART , contrée par l'école de Montpellier notamment sous l'impulsion de LONÇ qui a introduit les méthodes anglo y$@%$ saxonnes dans la description quantifiée de la végétation &$?* $:j ggF&g r.~..... 2~::. .?::

BUSH AND RELATED STRUCTURES

Quelques photos furent prises. mais les observations considérées insuffisamment documentées pour faire I'objet d'une note écrite.

Ce n'est qu'en juin juillet 1994 que je pus mettre en œuvre des moyens corrects mais personnels, en matière de déplacement et de navigation pour retourner sur les lieux et constater sur le site ponctuel retrouvé (grâce au point astronomique antérieur relocalisé par CPS) que je pus constater et documenter :

J la bande avait effectivement disparu sous sa forme originale par la totale mortalité des espèces annuelle et vivace du front de bande, et que la bande était encore repérable par des vestiges de Çrewia bicolor et quelques Boscia senegalemis. les premiers sec (et peut-être définitivement morts, car les pluies étaient tombées et aucun reverdissement ne put être constaté), le second actif.

J les interbandes en bordure du plateau et dans un moindre degré vers le nord est étaient effectivement, mais partiellement colonisées par à la fois du C. africana, mais aussi par des plages denses dlEuphorbia balsamifera, élément précédemment peu présent,. distribuées au hasard des ensablements superficiels.

J je décidai alors de consacrer du temps et des moyens à investiguer le phénomène. Une néo bande issue d'un ensablement fut étudiée en détail, photographiée au sol et marquée comme témoin.

J la conclusion principale est que les bandes se reconstituent, mais cette fois que le facteur déterminant est le vent porteur de sable qui crée des buttes d'ensemencement par les graminées. Celles-ci deviennent les germes en quelque sorte de nouvelles bandes. Ainsi je dispose actuellement d'un faisceau de présomptions quant à l'origine misemblable de ces bandes qui, dans un second temps, créent les différences de micro relief observées antérieurement ceci sur des surfaces sub horizontales.

*:* CONCLUSIONS

L'ensemble des observations quantifiées recueillies en 1994 doivent faire l'objet d'une publication générale sur les transformations de la végétation depuis le période des sécheresses de 73 et au delà. Pour être

PART l

TlGER BUSH AND RELATED STRUCT

pleinement convaincantes ces observations devraient être confortées par des photographies aériennes à diverses échelles.

Ce que l'on peut dire c'est que globalement cet exemple de I'évolution des BT montre que d'un point de vue de la perspective erronée de la DÉSERTIFICATION, FAUX CONCEPT, il n'y a pas lieu de lancer des programmes généralisés sur l'avancée prétendue du phénomène sans se référer à des études ponctuelles référencées par des observations précises sur l'état passé, dans la mesure où celle-ci ne sont pas seulement des descriptions plus ou moins lyriques de cet état. mais sont quantifiées.

L'ensablement loin d'être un facteur négatif peut très bien remobiliser l'eau de surface pour permettre la colonisation spontanée d'espaces précédemment nus, et ainsi remodeler le payssage de façon positive, c'est à dire dans le sens d'une plus généralisée colonisation par la végétation. Ceci est un phénomènbe d'anti désertification dont pesonne ne veut entndre parler. Et pour cause: la convention internationale de lutte contre la désertification ne veuzt regarder, pour des raison politiques et non scientifiques, que les aspceets négatifs.

+ B I B L I O G R A P H I E

AUDRY P. ROSSETTI, Ch. (1962)- Observations sur les sols et la végétation en Mauritanie du sud-est et sur la bordure adjacente du Mali (1959-1961) ronéo 267 pages FA0 Rome

CLOS-ARCEDUC. A. (1956)- Étude sur la photographies aériennes d'une formation végétale sahélienne. Bull IFAN XVIII, série A. no 3, 677-684

CLOS-ARCEDUC. A. (1 968)- La géométrie des associations végétales en zone aride Exploration aérienne et études intégrées, Actes de la Conf. de Toulouse p 41 9-421

ROSSETTI, Ch. (1996)- Réflexions sur l'utilisation des photographies aériennes pour l'étude du couvert végétal. Actes du Ile Symposium international de photo-interprétation Paris 1966 IV p 63-67

PART l

R BUSH AND RELATED STRUCTURES i;*--. ,-.Y,,%+ .>.̂ .$ ,,,, ,<* ;,:$&@&? , :;?$@$$: ,,,m

2.2X,,.& ..?$!szi; ?.U+XW ,..,, ~*

*:* SUMMARY

Arc or banded vegetation: static and dynamic approach examplified as from NE Mauretania

J This type of vegetation was shown to the author by Australian scientists in the Northern Territory in 1957 (Land Research Division CSIRO). The relation between horizontal structural distribution of that characteristic vegetation and micro morphologiy ha5 already been established at that time in Australia.

In 1960161 so called "brousses tigrées" with very similar image io those shown in Australia were discovered on maps and aerial photographs in NW Africa. actually in Central and Eastern North Mauretania At that time a french Geographer CLOS ARCDEDUC working on aerail photographs discovered similar pattern in Niger and elsewhere. Yet no satisfaactory detailed observations of the pattern werre amilable until this type of observations were made during an FAOlLlNESCO ecological survey of desert locust breeding sites in 1960)61.

J These observations fully confirmed the Australian hypothesis of local morphological factors being decisive(shallow slopes and small terraces where vegetation did concentrate). Detailed work is being presented from that period.

J The author did return to the precise site studied some thirthy years later in order to estsablish the dynamics of that vegetation after the great (1973-82) droughts. It was etablished that part of the pattern had disappeared and that a complex and different pattern had replaced it. The main difference between the previous situation and the present is pertaining to wind blown Sand on bare initially denuded interbands which now appear colonised by a low and open woodland.

J Precise sites of observations have been ermarked for further dynamical studies by GPS and one site has been mapped in detail. it is the strong view of the author that this type of observation can offer strong evidence for disputing the much too general view of overall desertification processes ending up negatively in vegetation degradation processes. lndicator function of arc or banded vegetation should be stressed in future.

The much to generalised view on desertification as proned by UN is here queried.

PART l

ased "Tiger pattern" generated m in Sokoto N o d t e m Njgeria

*

clear banded vegetation pattern can b aerial photogtaphs of the Northern part of Sokoto province in northern Nigeria. Becaus resemblance we called it after the th 1963164 'Tiger pattern" (FAO, 1966 1971, Sombroek G Zonneveld 1972)

this pattern appean to be quite different from a observed in Rep Niger and described by White 1971 "Bmusse tigrée". See also Thiery, d'Herbes and Vale summarize the genesis as moving vegetation arcs u sheet-run-off. We know banded patterns in wadi-fl

the latter ones. Chatacteristic are moving vegetate on more silty subsoils.

The banded pattern in Sokoto province however i caused by differences in surface hydrology due to variation in soi1 surface sealing that is connected with a by sheet erosion (pediplanisation) levelled former early Holocene dune landscape (Sangiwa covenand landscape). The difference in sealing is caused by a very small difference in si1 levelled former dunes and the filled in valleys.

The lower silt content in the valley filling is due to the resedimentation where, by water (and wind) action , the last remnants of silt that were still present during the Sangiwa period are removed, a feature well known fmm elswhere in cover sand formation.

' LTC. Enschede, the Netherlands PART l

ING THEORIES

The permeable soi1 of the former valleys bear a dense shrub. the stronger sealed remnants of the dunes show an open vegetation. Even relative big trees take part in, by the same process. filled in riven. with a vegetation that however is often disturbed by man and converted in grassland or temporary arable fields.

Ploughing in the sealed parts has no results because immediately after a rain shower a crust is formed.

The difference in genesis is clear from details of thi pattern. The banded pattern of the Sangiwa covenands shows up as a (pseudo-dune pattern with dunes and valleys perpendicular to the prevalling wind during the formation(about east-west).The so called "Brousse Tigrée" in valley bottoms shows bands that are perpendicular to the run-off direction that depends on the general relief of the area. In modelling the fint mainly aeolic forces of the past should be simulated, modelling the last one is a matter of run-off1 run-on simulation.

PART l

PATTERN AT A REGIONAL

anding patterns and associated vegetation structures appear readily on aerial photographs. Satellite data provide new methods to detect and map vegetation patterns at local and regional scales. Vegetation patterns depend on ecological pararneters as topography. surface

or soi1 features and rainfall. In western Niger, aerial photographs and SPOT satellite panchromatic and multispectral images (10 and 20 rn pixel size) have been used:

J to identify and classify the vegetation patterns;

J to extrapolate results at regional scale. Most patterns in the region cannot be discrirninated using Landsat TM imagery due to its too coarse resolution (30 rn).

The selected classification systern is OASIS. Initial data are supervised or unsupervised classifications, NDVl or color index thresholding. In a window of approx. 10- 15 pixels, each vegetation structure is defined by a reference pattern.. The composition of this window centered successively on each pixel of the image, is calculated and classified according to its similarity with the reference windows. We obtained thus a new classified image.

We propose a NDVl where red band is substituted by a combination of XS2 and panchromatic bands. This vegetation index has been validated along pattern transects in relationships with vegetation cover percentage. Other pararneters describing vegetation patterns include length and direction of bands or patches. Theses pararneters are extracted frorn image after binarization and reduction of vegetation

PART II

structures to a single line. The digital nurnber of each pixel represents the structure length or the direction angle. A photointerpretation on aerial photographs is used as the reference rnap. However, sorne limits are very difficult to delineate. The combination of autornatic classification based on bands length and rnodified NDVl is very accurate cornpared to the reference map. The extrapolation of results requires to verify the stability of windows composition for each pattern along a large latitudinal clirnatic transect. Multispectral data without panchromatic band gives also good results consistent with regional scale studies.

PART I I

processes of the herbaceous front

ylvie GALLE**, ristian VALENTIN**

ensely vegetated arcs parallel to the contour alternating with zones of bare soi1 are widespread in South West Niger. Seen frorn the air, this typical pattern resernbles a tiger fur. hence its cornrnon narne of tiger bush. Due to the occurrence of irnpervious soi1 crusts, the bare interband acts as the source of run-on for the.

downslope band. Therefore. the fringe between runoff and runon zones has been hypothetized as crucial for the dynarnics of the systern. This paper airns (i) to present a detailed characterisation of this peculiar ecotone. (ii) to infer sorne insights on the dynarnics of the tiger bush systern.

The zone of study was located 70 km East of Niamey and consisted of a rectangular of 60 rn long and 30 rn wide, perpendicular to a typical wavelength. The survey included rnicrotopography using a laser toporneter, surface features (crusts. termite rnounds, etc.). herbaceous and woody vegetation. Soil rnoisture (neutron probe) was rnonitored over four yean along a transect and related to the dynarnics of the herbaceous layer.

The pioneer zone is occupied by annual grasses (predorninantly Michrochla indica). and further down the slope by shrub (Cuiera Senegalensis) underlain by an herbaceous layer of Cyanotis lanata. The mean slope gradient (0.3[ across the bare interband) gradually decreases in the vegetated arc. However, the upslope herbaceous pioneer zone does not present a straight edge but local convexities (narned salients or capes) and concavities (pockets or bays). In the capes. the pioneer Cuiera senegalensis were rnost often associated

PART II ORSTOM. Niamey

** ORSTDM. Montpellier

IONING AT LOCAL SCALE

with micromounds which might be ascribed to increased soi1 porosity (root and faunal activity) and dust accumulation. In contrast, topography of the bay was flattened due to the accumulation of surface wash deposits as testified by the extension of sedimentation crusts. Furthermore, bays offered greater water resource availability than the capes. A minimum threshold water storage of 20 mm within the upper soi1 20 cm was determined to enable the colonisation by Michrocloa indica. However, the colonization the bare interband by this herbaceous layer was not constrained only by water availability because this threshold has been exceeded in the bare interband during most wet rainy seasons. Such limitation could be attributed rather to a low seed stock in the bare ground and to erosion and grave1 crusts which are much more resistant to seedling emergence than sedimentation crusts.

These results suggest that the contrast between capes and bays induce feedback processes. The microtopography of the capes tend to deviate the overland flow to the bays favouring thus the gradua1 levelling of these pockets by surface wash deposits, the water supply of the central zone of the arc and the colonisation of the bays by herbaceous species. This herbaceous layer tends in turn to attract termites and to prepare better environmental conditions for the establishment of pioncer shrub. Whenever these processes lead a bay to catch up with the line of the capes. these latter are turned into bays and the cycle can start again. This strongly support the hypothesis of upslope migration of the arcs, as a result of a preferential colonization of bays.

PART II

RUNOFF AND EROSION PROC

Gérard BELLIER*", Issa MALAM***,

anded vegetation patterning occurs in many arid and semi-arid regions of the world. Although its driving force may be the wind action in some areas, most authors hypothesized the role of the bare interbands as a souKe of runon for the downslope vegetated bands.

Such assumption is supported by the concentric distribution of arcs to the contour on the plateaux. Banded vegetation patterns occur on a very large variety of soils which have in common to develop impervious crusts when left exposed to direct impact of taindrops. Even though the role of these crusts has often been mentioned. no detailed study has been conducted on this issue. The objectives of this paper were to present: (i) the different types of crusts in this typical environment. (ii) their spatial distribution, (iii) some consequences its dynamic.

The study area is located in the Niamey region in Niger on lateritic plateaux. The methods of chatacterization included detailed field surveys at various scales. Samples of the soi1 surface were collected for mechanical analysis. thin section preparation. optic and SEM observations.

A wide variety of crusts were characterized. In the bare ground interbands, they included structural sieving. runoff. erosion. grave1 and sedimentation crusts. In the vegetated band. only a limited area was protected by a permanent litter cover and was therefore kept immune to surface crusting. In upslope and downslope edges of the vegetated band, structural and sedimentation crusts were common. A particular emphasis has been given to cryptogams which colonize surface crusts, more peculiarly erosion and sedimentation crusts, in the bare ground interband as well as under vegetation. Again a wide variety of

ORSTOM. B.P. 1141 6. Niamey, Niger ** ORSTOM, Bondy. France PART l *** Université d'Orléans

OFF AND EROSION PROCESSES

cryptogarns were identified.Due to a longer evolution, the crusts in the tiger bush environrnent were more cornplex than in cultivated land where they are rejuvenated by tillage practices. However, they involved a very clear succession along the transect of slaking, sorting, erosion, sedirnentation and bioturbation stages. This spatial distribution suggests a temporal succession and therefore an upslope migration of the systern.

PART III

RUNOFF AND EROSION PROC- $,< 8

* 2 &4 < Qi

ce water redistribution banded vegetation structures

mi CERDA, Anton IMESON,

i'";+**; ;<:~: - :ar.,, f&we;$,' .;;$;zTfj, ~'$<W&C&,"

: ~< *,*...-, " @$&-."*< anded patterns in soils and vegetation form par of L,~ 3;%;g$2$; ~ <+,;, ,. s$

,A"<"<? important discontinuities on semi-arid Mediterranean Q&L>?:; P$w .A:,<. ,<fyi*i,j slopes. At fine scales (< I m2) vegetation and well '+,-

structured soils are often located a t the outer rim of small terracettes. Together they form banded patterns.

distributed in a scattered way along contour lines. Runoff generated on bare soi1 surfaces in between the vegetation structures, infiltrates rapidly near the banded soi1 and vegetation structures (Bergkamp et al. 1995). This causes the structures to receive a supplement of water and dissolved nutrients. The supplement of water and nutrients is often thought to promote the dwelopment of the soi1 and vegetation structure and so to improve the structure's ability to capture water and : nutrient resources from adjacent bare areas.

The importance of the process of surface water redistribution for the water budget of the banded vegetation structure varies however dramatically from one site to an other. Cenerally, the importance can be expressed as the percentage of the water budget of the vegetation structures originating from surface water redistribution, in comparison to the rainfall amount reduced with the interception. The importance of surface water redistribution therefore depends on the infiltration envelopes of the bare surfaces, the rainfall intensity. the Storm duration and the recurrence intervals of specific wents.

For a large number of Mediterranean semi-natural sites in Spain, rainfall simulation experiments were carried out to determine infiltration envelopes. Sites were selected in Catalunya, Valencia and Cuenca on limestone. marls. schists and granite. 'The sites were located in different climatic regions.

* iandscape and Environmental Research Çroup, Uniwrsity of Amsterdam. Nleuwe Prinsengracht 130. 1018 VZ PART 111 Amsterdam, The Netherlands

FF AND EROSION PROCESSES

Data regarding rainfall intensity, Storm duration and recurrence intervals were analysed for al1 climatic regions.

Based on this analysis a complex interaction was found water redistribution for the water budget of the banded vegetation structure at fine spatial scale. (c I m2). A quantification will be given for different climatic regions and soi1 properties. It is concluded that the importance of surface water redistribution for the water budget of banded vegetation patterns varies dramatically between the various semi-arid regions with differences in both rainfall regime and soi1 types. Often its significance for the water budget of the banded soit and vegetation pattern is rather limited, depending on both soi1 surface as well as rainfall properties.

PART I I I

s of the vegetation striped patterns Mexico and theù influence

André MAUCHAMP**, Gera

he dynarnics of vegetation striped p

the whole systern.

*$. KEY WORDS

hydrodynarnics. interception, stemflow.

' Oatom, A.P. 17116596, quito. Equator " Lrtacion Cientifiea Charles Danuin. A.F? 17113891, quito, Equator. "' Cenid-Rarpa, A.P. Çomez Palacio, DÇO. Mexico

PARI' III

ing variations along a climatic ity and patterns of tiger bush

and Jean-Marc I'HERBES*

transect

RUNOFF AND EROSION PROC

anded patterns of vegetation have been recognised in many arid and semi-arid regions of the world, notably in West Africa where they are known as "tiger bush". They consist of bands of trees alternating with bands of bare soil. These vegetation stripes run parallel to the

contour on virtually flat to gently sloping surfaces, as the rnost effective intercepting arrangement of dispened runoff. The sequential pattern of runoff, interception, and run-on zones in this landscape constitutes a natural water harvesting systern. This pattern is expected to differ across space and tirne, as driven by global change forces like clirnate change and increased human pressure. The objective of this paper are (i) to analyse the variations of the tiger bush patterns as influenced by major factors. e.g. slope gradient and surface features along a clirnatic transect; (ii) to test a model of vegetation productivity based on surface water redistribution; (iii) to simulate the possible impact of global change in terrns of clirnate and land use.

The transect has been selected on a series of Panchrornatic SPOT images covering a 60 km x 200 km band from the vicinity of the W Came Park in southem to the region of Ouallam, near the Mali boarder. The mean annual rainfall ranges from 300 mm in the North to 700 mm in the South. Fifteen typical sites of tiger bush were identified on the satellite images and surveyed in the field. In each site. one to three transects were sampled including three bands and interbands. Major variables were siope gradient, soi1 surface conditions. e.g. soi1 crust types, litter, faunal activity, vegetation type. structure and density. The hydrological model used to sirnulate runoff from the bare bands and infiltration in the vegetation bands was based on the typology of surface features of Casenave and Valentin ( 1 992).

. . . . . . . . .. .. .

* ORSTOM. B.P 11416, Niamey, Niger P A R III

OFF AND EROSION PROCESSES

This study led to the identification of the combined climatic and topographic boundary conditions for tiger bush occurrence. For a given slope angle, dryer conditions result in narrowing of the vegetation band and a consecutive extension of the bare interband. The late stage sees the vegetation reduced to a series of aligned dashes increasingly surrounded by bare surfaces. Convenely, at the wet margin of the zone of occurrence, bare areas tend to be restricted to more or less circular crusted spots, often related to termite mounds. In a given location. patterns range from broad bands on nearly flat land (0.2%) to small "dashes" on steeper slopes (1.5%). Along the climatic transect crust types are also primarily controlled by slope gradient. Satisfactory relations were obtained between simulated infiltration in the vegetation band and assessed biomass, except where human disturbance was too severe. The difference between expected and assessed biomass indicated the importance of human use of the tiger bush. Simulations clearly showed that climatic change would have a minor impact on vegetation productivity compared to increased human pressure. Management strategies are proposed accounting for landscape hydrological processes as influenced by topographic. climatic and crust conditions.

PART III

RUNOFF AND EROSION

le cations within a patterned ground New South Wales, Aurtralia

D*, M.D. MELVILLE *, L. WHITE**

atterned ground in arid and semi arid Australia is landscape feature where spatial variation of vegetation and bare areas are rhythmically repeated. This patterning is exhibited in mulga tree communities. The study of the origins and characteristics of patterned ground in Australia has been generally confined to the

relationship between the hydrological regime and its control on plant growth. Our study shows the characteristics and proposes the processes controlling the soi1 cation distribution in chenopod patterned ground. There are important interrelationships between soi1 cations, hydraulic properties and the plant spatial variation across the patterned ground landscape.

The microtopography of patterned ground is generated by the "gilgai phenomena" as a result of the differential wetting and swelling of the clayey subsoil of the vegetated arcs. The distribution of soi1 cations within chenopod patterned ground is therefore not uniform across the landscape because the bare areas are "salt dumps". The bare area soils are dominated by sodium and it's absolute and relative concentrations decrease towards the centre of the vegetated arc. Magnesium and calcium have a similar elevated concentration under the bare ground but with greater concentrations relative to sodium under the vegetated arcs. Potassium is concentrated in the vegetated arcs and decreases under the bare soi1 surfaces.

The pattern of salt distribution is caused by the interaction of the differential runoff from the bare areas and enhance infiltration of water into the vegetated arcs. the differential evapotranspiration between bare and vegetated surfaces. the gilgai complex and the vegetation-induced

School of Çeography, University of New South Wales, Sydney, NSW 2052. ** CSIRO, Centre for Environmental Mechanics, ÇPO Box 826. Canberra. ACT 2601

PART III

OFF AND EROSION PROCESSES

Salt turnover. These factors cause the lateral rnovernent of salts out of the vegetated arcs and their concentration in the bare ground. We propose that overtirne the patterned ground and gilgai cornplex gets stronger and the Sal t distribution also becomes stronger until a point is reached where the systern is in equalibrium with the clirnatic environment. The implications for a recent land management practice are briefly discussed.

PART III

RUNOFF AND EROSION PROCE$& ,* &, 'F"

Australia

olyphase mosaic landscapes are widespread in the Australian arid and semi-arid zones. We describe here a mosaic developed in native grassland located south-east of Broken Hill in arid western NSW. The erosional processes active in the mosaic and the implications for broader hillslope erosion are the focus of this work.

The site typifies a regional low-gradient landscape and has an overall slope of 0.004 (approx. 0.2"). The landscape contains subtle topography related to occasional ephemeral stream channels and other small drainage features. The climate is arid, with the mean annual precipitation at Broken Hill, about 40 km distant, being 250 mm. Annual potential evaporation is about 2.5 m. Rain is distributed sporadically and unreliably throughout the year.

Vegetation at the study site is dominated by native grasses. This has a tussock growth habit, with 4-10 tussocks of varying size per square metre, and a tussock height of about 30 cm. The grass tussocks are concentrated into arcuate bands (mosaic components) that are separated by relatively unvegetated surfaces. Surface characteristics correspond closely to those described by Dunkerley and Brown (1 995) (J. Arid Env. 30: 41-55). The grassed bands display rough surfaces, pock-marked by crabhole collapse features. The unvegetated bands have a surface spread of small stones lying upon or embedded in the regolith surface. which often carries a patchy microphytic crust formed by cyanobacteria and lichens. No crabhole collapse features are dweloped within the unvegetated bands. The upslope margin of grassed bands is marked by a narrow zone of low forbs, while the downslope margin is sometimes associated with an increased abundance of chenopod

* Dept of Çeography and Environmental Science Monash University, Clayton Victoria 3168 Australia PAFI' III

OFF AND EROSION PROCESSES

shrubs. Average downslope widths of the mosaic components are as follows: unvegetated surfaces 9 m, zone of forbs 2.3 m, grassed bands 24 m. Widths of al1 components are quite variable. Grassed bands are associated with a fall of about 20 cm, while the unvegetated zones. being narrower and somewhat flatter, are associated with a lesser fall.

lntricate microrelief is associated with the surface mosaic. The unvegetated bands in general have a lower gradient than the grassed bands. but the gradient is not constant, the grassed bands being concave-upward in profile. A low ridge of material including coarse sand, and 1-2 cm high. is associated with the zone of forbs; this must be created by splash. perhaps aided by trapping of sediment in ephemeral litter dams lodged against plants. Downslope of the forbs. gradient increases steeply onto the uppermost parts of the grassed bands, a configuration favouring runon.

Hydrologic functioning of this landscape is akin to that described by Dunkerley and Brown (1995). The unvegetated surfaces shed water efficiently; wetting depths of only 2-3 cm were observed after a storm of moderate size. even after many hours of ponding. Water infiltrates more deeply in the zones of forbs and grasses, and was observed to reach 10 cm within the forbs and even more deeply within the grasses, from the same storm.

Erosional behaviour of the site is less clear. Sediment transport appears likely to be strongly compartmentalised as for runoff and runon. However, we have no data bearing on the issue of whether the study slope ever becomes an integrated hydrologie entity. transmitting water and sediment to the slope foot. We hope to investigate this aspect of site behaviour in a series of trials with simulated rainfall. Some aspects of the site bearing on erosional behaviour will be described in the paper, and their significance reviewed. These include the following:

O 5011 STRENGTH

Unconfined compressive strength was determined with a penetrometer a t many sites in grassed and unvegetated zones. The mean compressive strength is at least 5 times higher in the unvegetated zones. Splash of regolith particles from these sites is thus more difficult than from the more friable surfaces of the grassed zone. However. given that splash is enhanced by the presence of a thin water film, it may be that splash from the unvegetated zones is enhanced relative to that within the grasses, where ponding or surface flow seem highly unlikely, under

PART III

RUNOFF AND EROSION PROC

appropriate conditions. Shearing of particles from the surface by water flow will be restricted within the unvegetated surface owing to its mechanical strength.

0 S U R F A C E R O U G H N E S S

Surface roughness contrasts among mosaic components have not been reported previously. However, this aspect seems of great importance within runoff-runon systems. Surface elevation data were collected on a 10 cm x 10 cm grid with a falling-pin elevation probe in both sections of the mosaic. Roughness, assessed as the standard deviation of the elevation measurements, is nearly 3 times higher within the grassed bands (7.7 mm vs. 2.7 mm). The roughness of the grassed zones must retard any surface runoff, and increase the opportunity time for infiltration. Thus. regolith erosion from such sites must be restricted. The relative smoothness of the unvegetated zones must aid runoff movement there.

0 C A N O P Y C O V E R

This is significant within the grassed bands. By line transect methods, we have estimated this to be about 24%. Thus, some precipitation is intercepted within the grassed bands and this restricts the likelihood of runoff generation within these zones. Less than 10% cover was found in the unvegetated zones, relatively increasing the likelihood of runoff production there. Both tendencies are reinforced by the distribution of crabhole collapse features already mentioned, which must act to trap water within the grasses. Splashed grains entering the grassed zones must be partly trapped beneath the canopy.

0 S T O N E COVER

Stone cover is significantly greater on the unvegetated bands. Additional surface roughness is thus generated, and a fraction of the regolith fines lying beneath stones is protected from splash or entrainment by surface flow. Owing to the surface roughness, flow depths in surface runoff will be increased, potentially further restricting the surface forces generated by raindrop impact through the water-layer cushioning effect. Overall, at any site the behaviour of runoff will reflect a balance between the relative smoothness of the regolith surface and the drag exerted by the surface stones. Stone size, cover

PART III

OFF AND EROSION PROCESSES

fraction, and geometric arrangement thus become important aspects of site behaviour.

Overall then, the site conditions appear likely to restrict greatly integrated slope runoff and sediment entrainment. Degradation of the site characteristics by any disturbance seems likely to be associated with an increase in local erosion rates. However. much remains to be learned from mosaics in this region, which display some unusual features that will be described in the presented paper.

PART III

SOI1 WATER BA

ded vegetation pattern: h in Western Niger

1 EHRMANN**, Christophe PEUGEOT*

n intensive survey of a bare soillvegetation system was performed on a banded vegetation pattern (tiger bush) located on a lateritic plateau in western Niger from 1992 to 1994. The global functioning of the system is strongly related to the hydrological cycle, and therefore it is of concern to characterise and quantify its

components. The objectives of this paper is to study the spatial distribution of soi1 water within the bare soillvegetation system. and to establish the water budget at both the event and the annual time step.

Field measurements involved soi1 moisture (neutron probe) and local runoff (runoff plot) monitoring. They show that infiltration of water into the soi1 is strongly related to the location within the system: water stotage from the bare soi1 is low, much higher in the front bush, and very high in the centre bush. This spatial pattern is steady over yean. Runoff on the bare soi1 zone accounts for one half of the annual tainfall depth, and due the system geometry, the bare soi1 zone constitutes a natutal surface water feeding area for the bush that thus benefits of a water amount about 3 times higher than the annual tainfall depth. At the event time step, soi1 water stotage can exceed 12 times the tainfall depth in the centre bush. with an average value of 7 times. As shown on the water profiles. infiltration under the bush is tapid and deep (more than 6 m). probably due to a faunal macroporosity. Estimated to about one time the annual tainfall, deep infiltration is likely to contribute to aquifer recharge.

The water feeding of the bush by the bare soi1 zone, associated with vegetation dynamics allow to explain vegetation banding. The hydrological functioning describe here (runoff, sediment yield)

ORSTOM B.P 504534032 Montpellier Cedex 1 , France " ORSTOM, B.P 11416, Niamey, Niger

PAR'I' III

WATER BALANCE

associated with the issues of other recent studies. strengthens the hypothesis of upslope displacement of the vegetation bands over yean.

PART III

nce in the watershed of the river Çorouof- f of Burkina h o

he paper contains the exploration O

water balance in semi-arid regi study site is located in the watershed of the river Corouol (northern Burkina Fas region comes up with 462 mm at 375 mm at Markoye (long ter

rainfall can be diminished to 230 - 360 mm p pattern in that region is not developed in wel patterns. but is distributed in special landsca etc.). lnside these special landscape units contracted or diffuse. Soil properties are of natural vegetation and to agricultural crops properties are :

J Properties which affect soi1 moisture, s characteristics of silt-and clay-size particles, arrangement in structural units.

J Soil properties which affect the colloidal mineralogical character of the clay and its arran other particles.

J Characteristics of the regional hydrological system which affect the water properties of the soil, such as regional water-table levels and their season of variation.

J Soil properties. such as the presence of i layen, or profile heterogeneity, which affe root systems of the vegetation.

* lnstitute of Physical Çeography, University of Mannheim, L9, 1-2. 68131 Mannheim. Çermany PAR'I' III

WATER BALANCE

With the use of the "Catena" - principle three landscape profiles (each 3 km long) in the region of Gorom-Gorom. Saouga and Ménegou are mapped. The mapped landscape segments cross the different landscape units and inside of each landscape unit soil profiles (up to I to 3 mm depth), soi1 samples and soi1 analysis were carried out. Concerning to the classification of soi1 cover complexes (designated as "états de surface" by Casenave G Valentin, 1988) the soil moisture is measured by tensiometer in each landscape unit. The results of soi1 moisture measurements during two years (1 994 and 1995) are discussed. Additional to the mapping of soi1 types and soi1 moisture conditions the relief types and vegetation distribution in different landscape units were mapped. By the combination of relief, soi1 types and vegetation phytogeomorphologicaI units are derived. The phygeomorphological unites are transferred to patterns in aerial photos and spectral signatures of satellite images (e.g. Spot and Landsat TM).

The deduction of special spectral response patterns for the derived phytogeomorphologicaI units will be the next aim of the research. first results of these approaches will be demonstrated.

e3 REFEREMCE

CASENAVE, A ; VALENTIN. C, 1998. Les états de surface de la zone sahélienne - influence sur l'infiltration , ORSTOM. 1988.

PART III

SOI1 WATER

ect recharge to the deep unsaturated zone

M. EDMUNDS**, E. FELLMAN**, S.R. GAZE*, J. SUDLOW*

n estirnate of direct groundwater recharge below a region of natuml wood.land (tiger bush) bas been made in south West Niger using the solute profile technique. Data has been collected from a 77 m deep well dug within the study area covered by HAPEX-Sahel, an

rge scale energy. water and carbon balance experiment carried out during the surnmer of 1992. So far as the authors are aware this is the deepest profile to have been obtained in the Sahel region and as such provides a uniquely long record of recharge in the historic past. 20

The principle of the chloride profile technique is as follows. Chloride is introduced into the soi1 bath in rainfall and as dry deposition. Since chloride does not evaporate from the soi1 surface, and vegetation does not take up significant quantities, it becornes concentrated by evapotranspiration. Water in the soi1 can be broadly categorised into either upward or downward moving water, with the zero flux plane (ZFP), separating the two.'rhe upward water flux is driven by evaporation of water from the soi1 surface and transpiration through plants taking up water via their roots. Where there is sufficient infiltration for water to move below the root zone and the maximum depth of the ZFP, the water will continue to move down as deep drainage until it eventually reaches the water table.

Under conditions of recharge the ZFP represents the point below which a net, steady state, moisture and solute transfer takes place toward the water table. The arnount of chloride crossing the ZFP varies in relation to antecedent rainfall over one or more seasons. thus causing oscillations in the chloride profile. Under steady state conditions the

lmtitute of Hydrology. Wallingford. UK " British Çeological Sunrey. Wallingford. UK "' ICRIVn; B.P. 12404. Centre Sahélien, Niger "" ORSSDM, B.P. 1141 6. Niamey, Niger

R BALANCE

- average chloride concentration of pore water in this profile, Cs , will be - - - - proportional to the concentration factor, PI(P- , (where: P =

long term average precipitation and Ë = long tenn average evapotranspiration). This assumes no loss or gain of chloride to or from minerals, and that the water and chloride are transported at the same rate. Where the surface run-off component is negligible. and assurning there are no reactions with rninerals the recharge is given by:

Where:

- Rd is the space and time averaged direct recharge flux.

- - Fp and Fd are the average tainfall and net dry deposition fllmr (=input)

- is the averaged concentration of chloride in the rainfall

- Cs is the averaged interstitial water concentration

It follows that direct recharge can be calculated if the following values are known; the volume averaged concentration of chloride in the -

C - rainfall ( p ); the averaged interstitial water concentration (Cs ); the

long term average rainfall (F) and, the net dry deposition chloride flux - - ( F d ). If Fd = O. then the fraction of rainfall contributing to direct -

C C recharge is given by the ratio p I s .

The solute profile well is positioned a t latitude 13" 15'44" N and longitude 2" 3' 31" E, a t an elevation of 262 m. It lies a t the edge of a large area of tiger bush close to the village of N'douroua on the western side of the Say plateau. Locally, the ground slopes gently toward the south west a t a gradient of I in 120. The site is underlain

PART III

SOI1 WATER BA

by the Continental Terminal formation, a 49.4 m thick sequence of terrestrial sediments, which are Miocene to Pliocene in age. These sediments rest on a pre- Cambrian basement of mainly granitic gneiss. Only the upper unit of the Continental Terminal, the "Cr 8As argileux du moyen-Niger" is present in this area. The water table lies within the basement at a depth of 74. i m. which is 24.7m below the base of the Continental Terminal.

During well construction samples were taken from the unsaturated zone at the following intervals: every 25 cm from O - 10 m; every 50 cm from 10 - 62.5 m. then every metre to the bottom of the well. Pore water was extracted from each sample either by centrifugation or elutriation and analyzed for chloride; moisture contents of samples were obtained gravimetrically. This data has been used to produce depth profiles of pore water chloride concentration and moisture content throughout the unsaturated zone. From these profiles it has been possible to derive an estimate of historic direct recharge at the site. The chloride concentration of rainfall, which is required to make the estimate, was determined from the analysis of 123 rainfall samples collected from 5 EPSAT rain gauges in 1992. A mean recharge rate of 13 mmlyr is estimated for the upper 70 m of the profile. with a total residence time of 790 years. 'This is considered to be a representative estimate of the magnitude of direct recharge taking place below tiger bush areas.

PART III

veneta tion

WATER

site

rainage patterns of a watershed form the framework of important energy flow and nutrient cycles that occur on the landscape. If planning does not occur at a watershed level of resolution. activities on a smaller planning unit will be susceptible to being undercut by

events outside the project focus area. In southern Niger, banded vegetation patterning (known locally as brousse tigrée) occurs on laterite plateaus which are primarily used as communal grazing land and as a source of firewood. At the study site located near Hamdallaye, Niger (13'34 N, 2'35'E) these communal lands are severely degraded. This degraded portion of the landscape has become the source of runoff that triggers serious gully erosion and disrupts crop production downslope. Originally, the vegetation bands were 20 to 40 m wide. separated by 40 to 60 m wide sparsely vegetated corridors. The vegetation bands were dominated by 2 to 5 m tall deciduous trees. the most common species being Combretum micranthum, Combretum nigïicans and Acacia macrostachya: The edge of the tree bands were vegetated with herbaceous species such as Ctenium elegans and Pennisetum pedicellatum. Heavy firewood harvest and grazing over the last several decades has fragmented the banded vegetation allowing runoff to pass through the thicket gaps, wentually concentrating into torrents which cascade off the plateau.

Due to the vast scope of the degradation, i t was not practical to devise technologies to try to stabilise the entire plateau area at once. Therefore, technologies were tested that take advantage of knowledge regarding deterioration patterns and nutrient flows on the plateau that regulate the rehabilitation process. The emphasis of this approach was to concentrate on using, rather than combating, natural processes. It

Institut National de Recherche Agronomique du Niger. Niam~y. Niger " Texas A O hi University. College Station. Texas, U.S.A. "*Alabama A O M University, Normal. Alabama. U.3.A

PAKI' III

WATER BALANCE

was hypothesised that tree establishment and growth would be better in the gaps of degraded thicket bands than in the bare corridors between the bands. It was also hypothesised that if the gaps in the degraded thicket bands could be revegetated, the continuous thicket bands that grow on contour would once again act as natural barriers to overland flow and sediment transport and reduce downslope flooding and erosion. Further. it was hypothesised that runoff and eroded soi1 caught in the revegetated thicket gaps would accelerate the rejuvenation process.

The surface horizon of soils inside the thicket bands is a dark brown fine sandy loam, laminated layer that has strong coarse subangular blocky structure that parts to a thin platy structure. Surface soils in the bare corridors ands in the degraded thicket bands are severely crusted. whereas the soi1 surface in the thickets is not crusted. The soilllaterite boundary is undulating and abrupt, varying as much as 50 cm vertically over a I m horizontal distance. These poorly drained soils are rich in bases and available P. soils under the thickets have a relatively higher fertility than the soi1 of the barren corridors. The degraded portions of the thicket bands exhibit the relic fertility features of relatively recently vegetated soil. Soil of the bands are significantly deeper than those in the open corridors. Moreover, soi1 profiles in the vegetated and recently degraded portions of thickets exhibit extensive biological activity that is characterised by numerous biochannels and many roots. There is little evidence of biological activity throughout the soi1 profiles of the barren corridors between bands. This may partially explain the spatial arrangement of corridors and thicket bands.

Infiltration rates of the various components of the banded mosaic were determinated using a rainfall simulator to apply water at a rate of 150 mmlhr (equivalent to an intense rainstorm at the site) to I m2 plots. These tests were conducted before the onset of the rainy season (April- reflecting the hydrologic characteristics of the site at the onset of the rains) and midway through the rainy season uuly). Infiltration rates in the bare corridors and the degraded thicket gaps were low (32 mmlhr) and not significantly different between April and July. The infiltration rate within the thicket was significantly greater than in degraded areas, and was greater in April (77 mmlhr) than in July (41 mmlhr). Low infiltration rates in the bare corridors and the degraded thicket gaps were attributed to surface crusts and rapid sealing of the surface when the soils were wetted. Seasonal variations in infiltration rates within the thickets could be explained by changes in Iitter cover and plant biomass. During the dry season, deciduous trees shed their leaves resulting in a good surface mulch. The litter cover of the soi1 surface

PART III

SOI1 WATER BA

facilitates retention storage of runoff and protects the surface pore structure by dissipating the raindrop energy before reaching the soi1 surface. Litter biomes rapidly decomposed as the rainy season progressed due to enhanced termite and microbiological activities. Thus. the reduction in surface cover resulted in a decrease in infiltration rate.

Seedlings of Acacia holosericea and Prosopis juliflora were planted in 2m2 microcathments that were installed in both the degraded thicket gaps and the in the barren corridors. These microcatchments stored overland flow and accumulated soi1 and litter deposited by runoff events. The accumulation of soi1 and litter in the microcatchment harboured a large termite population which attacked and killed the P. juliflora seedlings. The A. holosericea seedlings seemed to be resistant to termite attack. The survival of this species was not different between the barren corridors and the thicket gaps. After 13 months the A. holosericea seedlings (IS cm tall when planted) had a 96% su~ iva l rate and had grown to an average height of 224 cm.

Flooding the gullies downslope has stopped as a result of tree establishment in the rnicrocathchments which re-established the continuity of the vegetation bands. Farmers downstream have recognised the practical value to them of restoring the banded vegetation on the plateau and have initiated additional collaboration with the Institut National de Recherche Agronomique du Niger to expand these reclamation activities to other degraded areas.

PARI' III

regulation in pattemed landscapes

$$ ky* a x,~; fi- $< ??- <

hilst it has long been recognised that soi1 fauna can t s , A 4 $gg~~:+~ play significant roles in mediating many fundamental

22 + ,m$< d-

ecosystem processes such as decomposition of organic matter, nutrient cycling. seed dispersal and herbivory, comprehensive studies of such activities are comparatively rare in many arid ad semi-arid

ecosystems. Ecological studies of soi1 fauna in semi-arid rangelands of Australia have been severely constrained by a basic lack of adequate inventory data on species distribution and abundance, especially for soi1 invertebrates. Furthermore. the rapid extinction of many of the smaller vertebrate fauna. some of which are now recognised as having had significant influences on some ecosystem processes prior to European pastoral settlement last century, has further hindered ecological insights into prehistoric landscape processes.

Nonetheless. increasing awareness in recent years of the importance of soi1 ecology and soi1 biology has seen several research projects undertaken throughout Australian rangelands. This paper will briefly review the results obtained from some of these by focusing on studies undertaken in patterned landscapes of semi-arid eastern Australia since vegetation patterning reflects fundamental patterning of habitats for a *

wide range of soi1 fauna. Particular emphasis will be placed on the ecology of different functional groups of soi1 fauna ranging from soi1 microarthropods (Acari) through termites, especially the widespread harvester termite Drepanotermes pemiger, to soil-inhabiting vertebrates. These will include such groups as the reptiles as well as marsupials like the now locally-extinct burrowing bettong (bettongia lesueur) together with exotic species such as the European rabbit (Oryctolagus cuniculus)

* National Rangelands Pmgram, CSlRO Division of Wildlife and Ecology, PO Box 84, Lyneham. AC72602 Australia

h -

The paper will refer to case studies undertaken in a patterned mulga (Acacia aneura) landscape near Louth in western New South Wales where initially, emphasis was directed towards examining, and quantifying, spatial patterns and abundances of surface-soi1 inhabiting microarthropods and surface features built by larger invertebrates, especially termites. Because these latter animals are regarded as key species mediating the decomposition of plant residues, both herbaceous and woody, field experiments have also examined the effects of termite removal on the rate of decomposition of senescent grass tussocks in two contrasting communities.

Because historical evidence suggests that episodic fires have had major impacts on ecosystem processes in the past. prescribed fire is now regarded as an important management tool, especially for controlling shrub encroachment, to be used whenever sufficient fuel stocks are available. There is little, if any, knowledge however, as to how different fire regimes have interacted with different functional groups of soi1 fauna. Accordingly, the effect of prescribed fire on the relative abundance and composition of soi1 microarthropods in different soi1 microsites will be discussed in the context of a patterned semi-arid mallee (Eucalyptus spp.) community.

Further discussion will centre on the hypothesis that some soil- inhabiting fauna such as the burrowing bettong required periodic fire to enable them to graze young foliage of shrubs resprouting at ground level, especially when there was little alternative green feed available. This fire x browsing interaction may in fact have constituted a prehistoric form of integrated shrub control which, in turn, reinforced spatial patterning of vegetation by impacting only on those patches capable of producing herbaceous fuel.

While macroscale patterning on a landscape scale ( 1 - 100 km) is controlled mainly by terrain factors, mesoscale patterning of vegetation, that is. 0.5- 1 km, is controlled instead primarily by biotic processes under the control of a suite of soi1 fauna which may also interact with abiotic landscape processes. This paper will discuss this hypothesis using examples from Australian research outlined above.

PART III

vegetation pattern

his study was "ndertaken in northern Burkina Faso, Yatenga province, in the Bidi site, as part of the SALT (Savanna on the Long Term) program. Using 137G to trace soi1 erosion and potential upslope migration of bandedvegetation in south-west Niger.

The climate is sudano-sahelian. The annual rainfall (400 to 600 mm) is concentrated over 3 months, from July to September. The study area is situated on the granitic substratum, with a very low slope (0.8%). Soils are tropical ferruginous, with a concretionary level close to the surface on the summit and lower slope. The vegehtion is distributed along the toposequence. from the hard-pan summit down to the bottom. Typical banded vegetation ("brousse tigrée") is located on the upper slope. It is characterized by vegetated (ligneous) bands. more or less parallel to the slope direction. with a herbaceous stratum (pioneer zone) upside and a degraded zone downside.

The aim of the study was to assess the distribution of termite nest populations (densities, surfaces and volumes involved) in this landscape and the effects on soi1 (texture. metabolism), on water fluxes (runoff, infiltration) as well as on vegetation (woody species, seedlings).

All the parameters were collected within a transect 2900 m long and 50 m wide: mound species, stage (alive or dead), height and diameter together with woody species (height) growing on mounds and outside. Termite foraging activity was followed along line-transects starting from selected nests. A special experimental device was installed to protect a vegetated band from upside runoff and to compare the evolution of the vegetation and termite activities with untouched bands.

'iaboratoire d'Ecologie, E.N.5.. 46 rue d'Ulm. 75230 Paris cedex 05 PART III

Termites are considered to be indicators of the evolution (up-slope migration hypothesis) of striped vegetation, as water fluxes influence the nest population structure over time. In turn. termites could exert a profound influence on the dynamics of the vegetation. through their actions on soils, water fluxes and nutriments.

Three main termite trophic groups were encountered on the study site: the humivorous group (Cubitermes). the gras-foraging group (Trinervitermes) and the fungus-growing group (Macrotermes). A high proportion of Trinervitermes mounds (86.4%) were dead and the densities are low as compared with other biotopes (maximum 25 nests ha- 1 , 2 live nests located within the upslope herbaceous zone). Cubitermes mounds could be locally abundant (80 nests ha-1). mostly within the ligneous band, 30.7 live nests in average ha- 1 , 26.8 dead ones). Macrotermes subhyalinus is relatively abundant in the striped vegetation: 21.2 live nests ha- I in the wooded zone. together with 36.8 ha- 1 dead nests. This last species is considered to be the most important in this biotope, as far as soi1 volume rehandled and surfaces are concerned. Therefore. most of the studies concentrated on the Macrotermes nests.

The present position of the eroded stages of Macrotermes nests (ranked from live nests to completely eroded). showed that none of the live nests are found outside the wooded area and that large numben of dead eroded mounds are found in the bare area. This observation could be interpreted by the shifting upslope movement of the wooded bands.

The mound bases represented 185 m2 ha-1 and the mound soil volume reached 19 m3 ha- 1 . Mound soi1 is richer in finer particles, mostly clay. There is a texture gradient in space (according to the structure and erosion of the nest): the mound center contains more than 50% clay. the mound wall contains 40% clay and the erosion cone 20-30%, as compared with less than 10% clay in control soil. As the mound dies and is eroded. mound material is further expended and enriches the surface soi1 with finer particles. Under heavy tain ("splash effect") and extreme climatic conditions, it is well known that soils rich in clay promote surface crust formation and therefore increase runoff. Lower infiltration rates and higher runoff production is seen on eroded material from Macrotermes nests. Ongoing infiltration experiments will confirm this fact and produce significant figures.

On the other hand. termite foraging activity, in mixing organic and mineral particles, enhanced the infiltration rates. In their foraging activities, located within more or less permanent territories, live

PART III

7- 7 9-7, < ~ige* ' y L,> . ':, %E% 4 ,,,, if*! g,g, 3 y@*#?$

$y&;:

.;-,2;~ >NW,+@J gp$.' 4 (>q* g' <. j /&&

Macrotermes nests built underground galleries, emerging on the surface <.a*;, "y: through foraging holes. The evolution of such foraging holes was - X

followed within transects across the biotopes and was seen to reach quite high densities within the wooded area: up to 240 holes per m2. Preliminary results indicated high activity within the upslope herbaceous zone. confirming the determinant role of termites in the infiltration within this pioneer zone. Also the overall termite foraging activity seemed higher in the protected band. as compared with the untouched one.

The influence of the soi1 rehandled by termite workers on soi1 metabolism (potential CO2 release under controlled temperature and humidity conditions in the laboratory) show a spatial (synchronie) and

y L S Y * ~ . a temporal (diachronic) gradient. In space, according to the structure *tm:~':

" ** of the mound and eroded material, the soi1 metabolism (carbon mineralization) is higher in the mound wall and in the eroded cone as compared with the inner nest on one hand (x 2 factor) and with the control soi1 (x 4 factor). on the other hand. Along with the erosion over time of abandoned nests. higher soi1 metabolism is recorded in soi1 from live nests and mounds in the early stages of erosion. while very old eroded mounds seemed to have lower soi1 metabolism activity than the control soil. However, this result has to be confirmed. If confirmed. it constitutes a further indication that old eroded mounds are unfavourable biotopes for vegetation establishment.

Sampling of the ligneous vegetation showed higher densities in mounds than outside: global density is 2248 ha-l as compared with 1303. The differences were more pronounced in bare areas. where most of the ligneous vegetation is concentrated on eroded Macrotermes mounds. Some species obviously preferred the moundsoil: Boscia senegalensis (density 71.2 per 1000 m2 of mound, 6.4 outside), Crewia bicolor (density 8.9 per 1000 m2 on mounds, 0.2 outside). The mound soil. which represented only 2.5% of the area, supported 57% of the Crewia bicolor populations, 22.2% of the Boscia senegalensis populations, 9.1 of the Acacia ataxacantha populations and 6.5 % of Pterocarpus lucens populations.

In conclusion. the data obtained on termite epigeous nests within this transect of banded vegetation pattern clearly demonstrated the role of this prominent component of the soi1 fauna in the structure and dynamics of the vegetation. The present-day distribution and evolution over time of termite nests could be utilized to confirm the upslope migration of the wooded areas. Termite action on soi1 rehandled (volume, texture, metabolism) and foraging activities throughout the

PART III

biotope (mainly in the wooded and herbaceous areas) influence the water and nutrient fluxes.

In summary, we could distinguish a "positive" effect on vegetation, as live nests increase the infiltration rates along a gradient starting from the mound and a "negative" effect as eroded material from nests increase runoff production. Under this hypothesis, the effect of termite action will be to reinforce other environmental conditions, such as climatic changes (drier periods) and human impact (overgrazing, fire- wood sampling). Ongoing further experiments are necessary to precisely assess the respective distances, intensities and variations over time of these "positive" and "negative" actions of termite populations.

PART III

POPULATIONS

Yatenga

n northern Yatenga banded vegetation is a rather scarce feature which appean closely linked to glacis (with a slope between 0.5 and 1.5%). Plains, with no consistent slope, bear a savanna-like vegetation with scattered trees and shrubs and a connected herbaceous

savannas). With regards to plains as well as glacis. soils are shallow and loamy above a more or less fragmented ironpan.

For both vegetation types. very few is known about their current dynamics and even life history traits of constitutive woody species are very poorly documented. The aim of this study is to draw information from an analysis of present spatial patterns of woody individuals. An emphasis is placed on the 2 dominant species of striped vegetation: Combretum micranthum C. Don, a shrub which averages 3 m in height at maturity and Pterocarpus lucens Lepr.. a small tree which can reach 8 m in height in favourable growing conditions.

Woody individuals were mapped for one reference plot in each vegetation type: the plot was 10.2 ha (320 m x 320 m) in the spotted savanna and 14.4 ha (300 m x 480 m) in the tiger bush. Locations of al1 individuals the height of which exceeds i .5 m were recorded. whatever the species; for some species, individuals between 0.5 and 1.5 m were also mapped.

Most data analysis was based upon the theory of point processes (as reviewed by Cressie, 1993); spatial patterns of woody individuals were studied using spatial pair correlation functions and ((Ripley's K- functions~, which allow testing of multiscale departures from complete

Ecole Nationale du Çénie Rural, des Eaux et des Forth .F! 5093 - 34033 Montpellier - France PART III

spatial randomness of individuals (null hypothesis); alternative hypotheses are attraction (clustering) or repulsion (regularity) between individuals.

The spotted savanna has a greater species richness and diversity than the tiger bush. As a consequence, the proportion of basal area accounted for by C. micranthum and P. lucens is far less (41 % as against 80%). Nevertheless, the latter species ranks fint in spotted savanna (29% of basal area) whilst the former is of less importance (12%). On the contraiy. C. micranthum is the dominant species in the tiger bush (46% of basal area). followed by P.lucens (34%). This last species is the only one that can be stated as playing a major role in both vegetation types.

In spotted savanna. most species display clustered spatial patterns (the most clustered are Combretum nigricans, Cuiera senegalensis and C.micranthum). Nevertheless, «mature» Pterocarpus lucens (height > 1.5 m) do not depart significantly from complete spatial randomness. whereas ((juvenilesn (height between 0.5 and 1.5 m) are clumped. The pattern of the tallest individuals (height > 6.5m) shows a slight repulsion under 10 m. and intraspecific competition might occur.

In tiger bush most species have highly clustered spatial patterns, except for mature P.lucens. Nevertheless, this species has stands slightly denser in vegetated bands than in interbands and is not completely independent of the overall structure. which is principally determined by C.micranthum. In some places. dense Pterocarpus stands are associated with poorly structured banding patterns. which might result from a taking over of Pterocarpus on C.micranthum.

The overall vegetation structures (spotted vs. banded) appear influenced by life traits of woody species. Due to its capacity for vegetative reproduction. C.micranthum is prone to constituting dense thickets, which can benefit from run-off along slopes, giving rise to or perpetuating banded structures (this species is thought to be dominant in extensive areas of western african tiger bushes). This advantage dwindles in the absence of systematic run-off, or under sufficient rainfall. In those conditions taller species, with efficient seed production and dispenal, have a greater abundance, yielding spotted patterns or poorly structured banded ones. This is particularly true for P.lucens but to some extent also for Crewia bicolor, Commiphora africana or Dalbergia melanoxylon.

PART III

POPULATIONS

iger bush is a particular vegetation pattern with densely vegetated bands alternating with bare areas. I t has been reported in many parts of the and and semi-arid zones. In Niger. bare areas serve as an impluvium for the downslope vegetated bands. The fint objective of this paper is to quantify the relationships between the

intraseasonal soi1 water flows and the seasonal vegetation dynamics (e.g. during a rainy season). We verified the need for the run-on from the bare zone to sustain the vegetated stripes. We focused on a zone where the most important gradient in soi1 water content and rapid changes in vegetation distribution are observed, i.e. the transition zone between the bare soi1 and the thicket.

Two representative vegetation stripes have been monitored from 1992 to 1995. Data were collected along transects placed perpendicularly to the vegetated bands, from the bare zone to the center of the vegetated thicket. along the line of maximum slope. In this way, the succession of vegetation features have causal relationships and connectance due to the flow of waterand the associated soi1 surface features such as crusts. Vegetation data include phenology and demography of the two main species of the herbaceous stratum from 1992 to 1994. and evolution of leaf water potential and phenology for the two main species of the woody stratum during the rainy seasons 1994 and 1995. Data on the crust distribution were collected once in 1992 and again in 1994. Soi1 moisture profiles were measured at depths of between I m to 5.6 m using a neutrons humidimeter (Solo) from 1991 to 1995. Pipes for use of the neutron probe were set on the same types of crust occurring from upslope to downslope on each vegetated band. A wall was built during the dry season 1992-93. It was placed just at the herbaceous

*Centre ORSTDM, B P 11416, Niamey, Niger **Centre ORSTOM. BP 5045 Montpelliez France PART III

-f;i'$!g" 2.*#

LATIONS DYNAMICS

stratum upslope boundary to stop the run-on from the bare zone to the vegetated band.

Data from 1992 allow us to give details of the processes involved and to verify the similarities in the functioning of the two stripes. The data from the following years allow us to compare the effects on the soi1 water and vegetation dynamics of deleting the run-on. Results show that for the herbaceous strata the distribution of Michrochloa indica (Linn.f.) P. Beauv., well adapted to an open environment, closely correlates to the distribution of sedimentation crusts. immediately upslope to the thicket cover. and Cyanotù lanata Benth, a shade- tolerant species, closely correlates to the distribution of microphytic crusts inside the thicket. Infiltration occurs first and deepest throughout the rainy season in the centre of the thicket under microphytic crusts. Herbaceous stratum grows simultaneously under the woody thicket. Then from the end of August infiltration increases upslope of the thicket, under the sedimentation crusts which were slightly colonised until this time, while only Microchloa indica moves upslope until there are insufficient seeds. The wall has no significant effect on the dynamics of the two main species of the low strata. in relation to the low differences in the soi1 water flows down to 40 cm. There are no significant consequences on the Combretum micranthum C. Don phenology and evolution of the leaf water potential because of the deep and extensive infiltration of the rains inside the thicket where this species is dominant and because of the run-on from the pioneer zone mainly covered by sedimentation crusts during the first part of the rainy season. On the other hand. the period and the extent of the reproduction phases of Çuiera senegalemis J.F. Cmel are greatly reduced and the leaf water potential increases very quickly after the last rains in relation to a great decrease in the soi1 water stock of the layers colonised by woody roots located in areas where there are sedimentation crusts and where this species is most abundant. Results are discussed in relation to the existing functioning models.

PART III

"@.$LQ . A

> A

SYNTHETIC MODELS OF FuNC'TIONEIG ,?y>;+"'

b % $< "

on resource îhe influence capture and productivity

;$#j,,y;" @fg&g$ >;::.,. :L>x* .:;$$&;&fi -- $.&+4;1;

n rnany serni-arid environrnents around the world, -3:; , &;$$:y; ;JF&:2@

lirnited iainfall and runoff-runon processes lead to ,&j;.iz4g$ t"nw qqFmx; .-&: .&y ~%,;,u, vegetation patchiness. This phenornenon has been ,%$ ,,.$,,><",

described as two-phase rnosaics (e.g. Archer 1990; ;&$$*37ir?$; &#*:& Montana 1992). banding (e.g. Belsky 1989). stripes :wM8 (e.g., Cornet et al. 1992). tiger bush (e.g. Thiery et al. #@v#

1995) and grove-intergrove (e.g. Tongway & Ludwig 1990). This i$,&&$j; landscape patchiness functions to optimise plant production by

g$3gqgj;. @$,F .... " :2j$$

concentrating and consetving limited water and nutrient resources >t@$gl!$: ;Sa,, % :+&:

(Ludwig & Tongway 1995). This optimisation is based on the theory $g:%f!j$ that arid lands function as source-sink or runoff-runon systems (Noy- g@@: Meir 1973). This theory predicts that in environrnents with lirnited ggg@ resources, plant productivity will be higher if these resources are g;zb;r3p2:, ~.a:p:+r;sa ., Id-Ym.z7

concentrated into patches and not uniformly dispened over the , ,,< > * ,. ..,

landscape. The question addressed in this study is: how much does the type of patchiness influence loss of resources and hence productivity compared to a landscape with no patchiness? Does it really matter whether patches are small and scattered (stippled) over the landscape, or occur in short bands (stripes) or as long bands (strands)?

A 'flow-filter' simulation rnodel was developed to quantify how runoff flows down, and possibly out of a landscape (Ludwig et al. i 994). If the amount and intensity of iainfall (R) exceeds the water infiltration rate (IR) or water storage capacity (SC) of the soi1 within the interpatch area then runoff (ROff) occun. This ROff can run out (ROut) of the landscape system or be captured by patches. If the IR and SC of the patch is exceeded then ROut occun frorn the patch. The total ROut frorn a landscape following a iainfall event a t time (t) is also a function of the slope (S), total area of patch (AP) and area of interpatch (Al):

ROut t = f (R,IR.SC.S,AP,Al)t

* CSlRO Division of Wildlife & Ecology, PO Box 84, Lyneham, ACT 2602, Australia PART III

ETIC MODELS OF FUNCTIONING

Annual net plant production (NPP) for the landscape system was estimated by using a linked simulation model, SEESAW, which sirnulates the ecology and economics of serni-arid woodlands (Ludwig et al. 1992, 1994). A subrnodel within SEESAW computes annual net primary production (NPP) through time (t) as a function of plant available rnoisture (PAM) and available nutrients (AN) and temperature (TEMP): NPPt = f (PAM,AN,TEMP)t

PAM is estimated by a submodel, called WATDYN, which cornputes soi1 water balance dynarnics (Walker and Langridge 1995). For the simulations we used a serni-arid landscape system of fixed size (I ha), slope (1 %) and shape (rectangular), but with different patch structures:

J stippled - small, circular and dispened patches;

J stripes - larger, elongated and clustered patches;

J strands - long linear and basal patches; and

J no patches. Patches occupied 25% of the area (or were absent).

Pararneter values used for the simulated semi-arid landscape were based on actual field rneasurernents. The rainfall and temperature data used to "drive" the simulations was based on a 31.5 yr (mid- 1962- 1994) from Cobar, NSW.

The loss of runoff frorn the sirnulated landscape (Runout) was greatest frorn the systern with no patches. as expected, while the stripe and strand banded patterns were rnost efficient at capturing runoff (Fig. 1 ) . The stippled pattern was interrnediate because runoff frorn rains could be lost out of the systern by flowing between and around srnall patches. Because striped and strand patterns captured and conserved the-most rainwater, these systerns also had the greatest net prirnary production (Fig. 1). whereas the systern with no patches had the lowest NPP, about half that for stripes and strands.

These simulations dernonstrate that the presence and patterning of patchiness can have significant influences on the potential prirnary productivity of serni-arid landscapes. If land degradation caused a loss of patchiness, hence a loss of resources such as rainwater, sedirnents and nutrients. which are already very lirnited in supply in these landscapes, significant declines in plant production can be expected. It is known that land degradation often leads to a loss of landscape patches (Tongway and Ludwig 1994). and semi-arid landscapes used as rangelands for grazing rnust be rnanaged wisely to avoid degradation.

PART III

SYNTHETIC MODELS OF FUNCTlONl

Fig. 1. Runout and net primary production (NPP) from o simuloted seminnd landscope for three potchtypes ond no potches.

R E F E R E N C E S

ARCHER, S. 1990. Journal of Biogeography. 17: 453-462.

BEUKY, A. J. 1989. Journal of Arid Environments. 17: 265-270.

CORNET A. et al. 1992. pp. 327-345. Ecological Studies 92, Springer Verlag. NY.

LLIDWIC, J. A. G TONCWAY. D. J. 1995. Landscape Ecology. 10: 51 - 63.

LUDWIC, J. A. et al. 1992. Mathematics G Cornputers in Simulation. 33: 373-378.

LUDWIC. J. A. et al. 1994. Pacific Conservation Biology. 1 : 209-213.

MONTANA, C. 1992. Journal of Ecology. 80: 3 15-327.

IVOY-MEIR, 1. 1973. Annual Review of Ecology and Systernatics. 4: 25- 51.

THIERY, J. M. et al. 1995. Journal of Ecology. 83: 497-507.

PAKI III

HETIC MODELS OF FUNCTIONING

TONCWAY, D. 1. G LUDWIC, 1. A. 1990. Australian Journal of E c o I o ~ ~ . 15: 23-34.

TONCWAY, D. 1. G LUDWIC. ]. A. 1994. Pacific Conservation Biology. 1 : 201 -208.

WALKER. B. H. G LANCRIDCE,]. L. 1995. Ecological Modelling. 73: (in press).

PART III

of pattern

,FEVER

formation

w@ w "?$ g$

U+

SYNTHETIC MODELS OF FUNCTIOI$~~~~ * , l,

3ki v* g9 +tir$$;

e propose a mean field description of vegetation population dynamics which takes into account the following processes:

J Cooperative vegetation reproduction: cooperativity represents the favorable influence of vegetation on

environmental conditions which in turn enhance reproduction; >'$$y$,*

J Inhibition due to competition of vegetation for space and other ";$!>$ k &.am' [$&p,$x

environmental resources;

J Death andlor destruction determining the average life time of vegetation.

The fint two processes are non-local and therefore each one is associated with a specific spatial range. The third one is, by comparison, essentially local.

We show that the formation of vegetative patterns is due to a symmetry-breaking instability, the occurrence of which is critically determined by the ratio of the reproduction and inhibition spatial ranges. The characteristic wavelength of the pattern is determined only '4

by the parameten of the dynamics.

We study the appearance of patterns for isotropie and anisotropic (ground slope. prevailing wind direction ...) environmental conditions. In the latter case, we discuss the mechanisms which orient the vegetation stripes either parallel or orthogonal to the direction of anisotropy. We account for the emergence of arc-shaped patterns.

PART III

amic on Tiger Bush in Niger

nd Luc ABBADIE*

rousse tigree in Niger is made of vegetation arcs alternating with bands of bare ground. Many studies have been conducted in this banded vegetation type in order to understand its dynamics. Data are available about waterflow and competitive relationships between

plants, and have been recently used to elaborate some simulation models.

Up to date, few data were available on soil organic matter storage and mineralisation, two major processes controlling soi1 fertility. However, nutrients availability could constrain both the rate and the pattern of the plant communities succession. The purpose of this study was to document the origin, the distribution. and the quantity of organic matter in soil, and the dynamics of microbial activity (carbon respiration and nitrogen mineralisation).

Soil samples were collected in April 1995 during dry season at 60 km NE of Niamey (Niger, West Africa). The study area was situated exclusively on shallow gravely soils of plateau lands formed by a thick laterite cuiras of Tertiary age and on gentle slopes.

Soil texture analysis showed that soi1 samples contained high proportion of clay both inside the vegetation arc and the bare area, with a high variability inside each area. The high clay content measured upslope in the bare area could be due to a past termite building activity. As termites only live inside the vegetation arc, these high clay contents support the hypothesis of a moving of vegetation arcs.

* idoratoire d'écologie, 36 rue d'Ulm, 75230 Paris cedx OS. Frcmce PART III

13C natural abundance analysis showed a high contribution of C3 plants to soi1 organic matter storage inside the vegetation arc. Nevertheless, the contribution of C4 plants was not negligible inside the living trees and dead trees areas. The highest d 13C were not measured inside the grass area, but just upslope, in the settling zone. This indicates that grass roots extend far beyond the grass area. lnside the bare area. the natural abundance of 13C increased from upslope to downslope.

Total organic carbon and nitrogen contents were very different between the bare ground area and the vegetation arc. No significant variations were observed inside the bare ground zone. On the contrary, a high variability occurred inside the vegetated area: the highest C and N concentrations were measured in the living and dead tree zones. and the lowest in the settling zone. Total carbon and nitrogen contents were not related to clay: clearly. organic matter pools were not dependent on soi1 texture, but on quality and quantity of dead vegetal matter entering into soil. The structure and the nature of the vegetation are the major factors driving soil organic matter dynamics. C:N ratio did not vary significantly neither inside the vegetation arc nor the bare ground zone.

A linear correlation between total organic carbon content and carbon dioxide production during incubations under optimal conditions showed that quality of soi1 organic matter did not vary with zones: the intensity of potential soi1 respiration was only dependent on the quantity of organic carbon. Potential respiration was not negligible in the bare area soi1 samples, suggesting the input of fresh organic matter, perhaps resulting from roots exploiting this bare ground area.

Tiger Bush is a variable ecosystem in space and time. due to soil physical structure, likely controlled by past and present termite activity, and to vegetation dynamics. This heterogeneity induces a high variability of soi1 organic matter storage and mineralisation, and of nutrients production and availability.

In spite of the relative weakness of soi1 organic matter pool, the environment was relatively "rich". This richness was both depending on fluxes of nutrients and on the pattern of space exploration by vegetation. Microbial activity and 613C measurements inside the bare area allow us hypothesise that soi1 organic matter and nutrients dynamic inside the vegetation arcs are partly dependent on the bare area functioning.

PART III

GENESIS OF BANDING GENERAL

slope migration of banded vegetation

hristian VALENTIN**, A. WARREN nd Jean-Marc ~'HERBÈS**

"Y-? ecent studies of banded vegetation have suggested a ''!::z2 successioml model, in which bare bands are colonised ;&>y

by a pioneer front on the upslope side of a vegetation ;Gzx,;3 band. Vegetation patterns in south-west Niger have %2s:$; :;e%& been interpreted to suggest that spatial transitions

+vm

reflect this form of temporal succession. and in these patterns there is !@$& m*

corroborating evidence for slow up-slope migration. However, given the inherent difficulties of long-term field experiments there are few data to $kE$!$ judge the validity of this model. >?W"< a.i&gt

The use of the artificial radionuclide caesium- 137 ( 137G) to provide information on net soi1 flux over the past 30 years offers potential in this regard. Furthermore. the identification of various types of soi1 crust. which can induce different types of hydrological behaviour. provides valuable information for predicting soi1 evolution. To test the hypothesis that banded vegetation migrates upslope. a 70 m transect encompassing two vegetation bands and a single bare lane was sampled in south-west Niger. The transect was aligned orthogonal to the bands and approximately parallel to the direction of water, soi1 and nutrient flow. Soil samples for gamma-ray spectrometsy and particle-size analysis were collected along the transect a t twenty one locations with I m intervals in the lower part and three samples were obtained on the upper part. Prior to collection, the soi1 surface characteristics were examined to distinguish between crust types and to identify the presence of termite . activity.

The results emphasise the utility of these techniques for examining the net redistribution of soi1 over periods of three decades and its relations

Department of Physical Çeography, University of Lund, Box 118 5-221 00 Lund, Sweden " ORSTOM, B.P 11416. Niamey. Niger PAFI III "' Department of Çeography, Unrversrty Collegeiondon, 26 Bedford Way. London. WClH OA?? UK "" Department of Physrcs and Astronomy. University College London, Çower Street. London WClE 6B7: UK

SIS OF BANDING GENERAL MODELS

to vegetation succession. The amount of soi1 eroded was found generally to decrease downslope, whilst the proportion of fine silt in the soi1 generally increased downslope. These patterns correspond with the location of the erosion and sedimentation crusts identified using a standardised classification. The relations between microtopography and net soi1 flux may also explain some of the spatial variation in soi1 redistribution processes. The intensity of crust and 137Cs measurements on the upslope edge of the lower vegetation band enabled the calculation of the upslope migration rate (c. 0.3 m yr-1) which coincided with other independent studies in the same region. Moreover. because considerable spatial variations in topography and soi1 flux were found to occur over very small distances, further detailed studies over larger areas will be needed.

PART III

moue upwards in Niger

Jean-Marc ~ ' H E R B È S , Aboubacar ICHAOU

he dynamic of vegetated arcs in a banded vegetation on plateaux in west Niger is evaluated by multidate aerial photographs. Two methods were used: aerial photographs and dendrochronology.

Aerial photographs are acquired in 1950 (scale = 1 :50000). in 1975 (1 :60 000) and 1992 (1 :50 000 and 1 :25 000). 25 et 42 years later. Previous estimations of the vegetation shift by others methods gave velocity of 4 meters for 40 years. The scanning of aerial photographs provides a pixel size between 2 and 5 meters on ground. Bands are 20 - 30 meten large. i.e. 4 - 15 pixels. Geometric corrections for

superposition is a prerequisite to obtain a good estimation of spatial dynamic. 1950 photographs are used as references. Localization error for other dates is less than one pixel. Three small plateaux are selected. Their surface don't exceed I to 1.5 km2 with a pattern of 6 or 7 bands or more. We generated a classified image with one class of bare soils and two classes of vegetation : high and low coverage by bush and herbaceous mixed. Some confusions remain with gravely surfaces and degraded vegetation bands. The images of the three periods are superimposed and the changes of each initial class to bare or vegetated are studied in 1975 and 1992.

From 1950 to 1975. the first effects of drought since 1970 involved a decrease of the vegetation cover. Human degradation remains limited on these plateaux during this period. We observed also an overall contraction and a smoothing of the vegetation stripe boundaries. The bands are frequently disconnected. On the other hand. some evidences of upslope dynarnic of vegetated bands are detected as :

PARII' III

SIS OF BANDING GENERAL MODELS

J recent connections between vegetation bands,

J closing of small bays along the bands,

J thickening of bands. These processes are always directed upslope and the vegetation encroachment ranges between 5 and 20 meters. Convenely, band thinning is mainly restricted to the opposite downslope side that suffers insufficient water runon. From 1975 to 1992, the rainfalls are always below average and the system tends to a new equilibrium with a general thinning of vegetation bands mainly downslope. Howwer, some upslope colonization of bare surfaces and at the end of vegetation bands still occur.

Shrubs and trees datation were performed along a transect 20 meters large and 400 meters long. across 6 vegetated arcs. Mean ages were determined for three zones within the arcs:

J upslope or pioneer front,

J close bush or centre zone and

J downslope or degraded zone. These ages are respectively 6 years (ma IO), 10-13 (max 41) and 14-15 (ma 26). One can thus infer fint that the whole band is no more than 40 yean old, secondly that an age gradient does exist within the band. Such data might be also interpreted through a whole band migration within less than 50 yean. However, such assumption is not supported by the study on aerial photographs and with other temporal indicaton of band migration. Rather. such data are interpreted as resulting from interna1 vegetation turnover within the bands related to life history and life cycle of each species involved.

Due to the relatively short timespan and drought impacts, upslope migration of vegetation bands was not systematically demonstrated in every part of the studied area. Howwer. a partial validation of the model predicting an upslope migration of the banded system (Thiéry et al.. 1995) was obtained.

PART III

GENESIS OF BANDING GENERAL

terning in a subantaretic forest of outcome of the interaction

S*, F. GALLART**, O. BIANCIOT G. del BARRIO*

E* ,255" i @

anded structures have been investigated in a <

Nothofagus betuloides primeval forest from Bahia del tee: Buen Suceso. in the eastern corner of Tierra del Fuego island. growing on spodosols developed upon silicic shales. in a cold oceanic climate. with 5? C annual

temperature and 600 mm rainfall. Bands are oriented perpendicular to wind direction. with older and dying trees in the windward edge and a seedling regrowth in the lee side of each band. Forest structure, species composition and relevant soi1 properties were sampled in transects across the banding, along a hillslope gradient and in a non disturbed stand. Results show that wind causes about a 50% reduction of maximal ages and sizes of the larger trees, and that stand growth processes, such as self thinning. basal area and high increases, specific composition of the undentorey occur in windward direction. as well as changes in soi1 properties such as CIN ratio and redox potential increases.

Field results allowed to raise a hypothesis to explain how wind is able to generate this pattern. 'The core of this hypothesis is that bands develop when vulnerability of trees to wind damage increases with age, and with lack of protection from older windward trees. In such conditions, bands are the outcome of a tuning between rates of tree growth and of the leeward movement of the dead front.

A simulation model was built up, based on this hypothesis and field observations. The model works on cells of 50 m2 and starts with an heterogeneous forest of random distributed tree ages. In each cell. trees are affected by an age dependent exponential probability of dead which

Gtaeion Experimemental de Zonas Aridas (CSIC). Çeneral Segura 1. 04001 Almeria (Spain) ** Instituto de Ciencias de la Tierm ]aume Almem (CSIC), Marti i Franqu{s sln. 08028 Barcelona (Spain) P A R III *** CADICICONICET: Ushuaia (Argentina). **** IFONA, Ushuaia (Argentina)

SIS OF BANDING GENERAL MODELS

increases when they are unprotected by older cells in windward direction. Model outputs were successfully tested with reality using field sarnpling and rnaps derived from aerial photographs taken from helicopter. Sirnulated bands are like those observed in the field, they require a fetch of around 20 cells and a nurnber of iterations equivalent to two tirnes the maximal age of trees, band width decreases with wind intensity and at large values of the latter. the pattern becornes disrupted and twirled structures appear. This interpretation of banded patterning assumes a tuning between a directional disturbance and an oscillatory process, such as stand regeneration, growth and decay.

A discussion is added cornparing this type of pattern with othen that are generated by the interaction between plant growth and sedirnent rnovernent in subalpine and serni-arid hillslopes covered with grasses such as Festuca eskya or Stipa tenacissima.

PART III

ample sites of "brousse tigrée" and related vegetation types are described for central and eastern Mali. Species composition and physical structure of the herbaceous layer as well as woody plant population were recorded at al1 sites together with soi1 and resource management variables. Herbage yield was measured whereas foliage

yield and wood mass were calculated using calibrated allometric relations established. for each species. between yield or mass and the basal circumference of the stems (Cissé et Sacko. i 987). "Brousse tigrée" differs from other sahelian vegetation types by its extreme patchiness and composition largely dominated by woody plants. It is characterized by the regular alternance of bare-soi1 stripes with dense g linear thickets arranged perpendiculariy to the slope. It develops on poorly permeable soils provided they are located on uniform and very gentle slopes so that bare stripes act as impluvium for the downslope thicket. Soils are often composed of a thin silty horizon that directly overlays the bedrock or an indurated pan which may surface in bare stripes. Facies of "brousse tigrée" have been distinguished on the base of the width and shape of the thickets and bare stripes (Ambouta, 1984) as well as dominant woody species (Hiernaux et al.. 1983). Facies types are related to the thickness of the top horizon. its texture, the nature and permeability of the underlaying material, the steepness of the slope and the average rainfall. In some of the more arid sites, depending on the orientation of the thicket relative to the dominant winds, Sand micro-dunes may be distributed across the bare stripes. or Sand may accumulate on the windward side of the thicket (Leprun, 1979). The presence and type of Sand deposit contributes to further diversify "brousse tigrée" facies. It also alter the spatial redistribution of

Paper presented at the h m m a on b n g Jërm (SALT) htemtiond Symparilm on Tmded wgetation patterning in arid and m i u r i d envimnment ccological pmcessa and m q u e n c e s for management: 2-5 April 1996. tùris, Frcmee. PART IV ** Range eeologist of the researdi teum of hternatiomi Livcstock Rrsearch Institute (ILRI) at ICRIYITSahelian Centre. B.P 12404, Nimney. Nigec

UCTlVlTY OF PATiERNED VEGETATION

water and nutrients. in ways that sometimes lead to a collapse of the "brousse tigrée" system as observed in the some of the more arid sites.

Quantitative vegetation structure and yield variables are compared between sites. especially between neighbor sites that differed by the degree of vegetation patchiness. There were no clear superiority in the overall site yields of the 'brousse tigrée' when compared to sister sites with diffuse vegetation. However, the patchiness of the 'brousse tigrée' affected species composition with the woody-plant mass being enhanced to the detriment of the herbaceous layer yield. On the other hand. the species diversity of the herbaceous layer was higher in the 'brousse tigrée' because of the wider array of edaphic niches that resulted from spatial redistribution and local concentration of water resources and shade.

The high diversity of the botanical composition relative to other sahelian biotopes was also observed in the herbaceous layer of five 'brousse tigrée' and related vegetation sites that were monitored from 1984 to 1993 among 24 rangeland sites in eastern Mali (Courma). The highest spatial heterogeneity of the herbaceous layer was found in 'brousse tigrée' where coefficient of variation of mean herbage mass averaged 360% over 10 years compared to 88% on sandy and clay soils (Hiernaux, 1995). High spatial heterogeneity and species divenity of the 'brousse tigrée' did not attenuate the interannual variation in herbage yield despite much lower average yields. 286 (s.e.76) kg ha-l , than on sandy soils, 961 (128). and on clay soils, 1448 (264) kg ha-1. The coefficient of variation of the IO-yean mean yield averaged 118% for the 'brousses tigrées' located in the more arid zone (200-300 mm annual rainfall) and 72% for the less arid sites (300 to 400 mm annual rainfall). This compares with 88% and 50% for sandy sites located in each zone, respectively. and 63% for the five clay soi1 sites across the zones. Vegetation patchiness did not impede a severe degradation of the vegetation communities that occurred in the three more arid sites. Following two decades of drought, the herbaceous layer was the fint to diminish although species diversity was maintained in a few refuge niches. Then, as the dry condition persisted. most of the woody plants died with the exception of Boscia senegalensis and Maerua crassifolia shrubs. A few species such as Acacia erhenbergiana and Commiphora africana for the woody plants, Microchloa indica, 77agus berteronianus and Schoenefeldia gracilis for the grasses recolonized afterwards. However, these new comen spread in linear patterns along the network of gullies that resulted from the change of sheet run-off to concentrated run-off. This major change in the spatial pattern and redistribution of

PART IV

PRODUCTIVITY OF PATTERNED VEGE

water and mineral resources leaves little hope for a natural regeneration of the "brousse tigrée" at these sites.

Crazing by livestock and wood cutting by humans are often implicated in the degradation of the "brousse tigrée" but other authon consider that they could be the origin of the patchy pattern of the "brousse tigrée". Selective and intensive grazing of herbs during the growing season may, indeed, favor the dominance of woody plant. in addition, the patchiness of the vegetation cover reduces the risk of fire which could have balanced the influence of grazing in the competition between herbs and woody plants. Except for species composition of the herb layer, little widence was found of an influence of grazing on the vegetation structure and yield along grazing pressure gradients established from livestock concentration points such as water points, cattle path or resting sites uurner. 1992). Firewood production in the Sahel consist mainly in harvesting dead wood (Benjaminsen, 1993) which has limited impact on the ecosystem. However, repeated cutting of live wood, chiefly as building and fencing material, contributes to the gradua1 loss in plant density and divenity. The increasing practice of thicket clearing for sorghum crop, in southern Sahel, and cropping millet on the Sand deposits, in northern Sahel, affect the spatial redistribution of water on which the system is based. The concentration of the run-off and the mobilization of the Sand deposits rapidly lead to severe soi1 erosion and threaten the resilience of the "brousse tigrée".

PART IV

TOWARDS AN IMPROVED LAND MANAGEMENT IN ME ARlD AN

tation pattern nment and fire in Spain

ERART and Anton IMESON*

ata are presented from two locations Spain where patterned or banded ve on abandoned land or where vege from wildfire. In both cases patterns investigated as process-pattern pheno aim of understanding how different

environmental gradients influence pattern evolution.

On abandoned land, patterns occur a t different scal scale there are areas where Plantago.spec germinates influences the accumulation of silty material. At the form elongated steps that create a characteristic mi the patch and slope scale Stipa t. tussocks form an hexagonal pattern on level areas where water infiltrates in and around the tussocks. On sloping areas the Stipa tussocks form parallel ovoid bands. They intercept fine and coarse material being eroded on the slopes by both overland flow and the hooves of sheep and goats. This also creates a distinctive micro-topography.

Rainfall simulation experiments were undertaken in combination with monitoring activities in order to investigate the effects that key- processes of sediment and water movement have on the patterns. Other methods include controlled experiments and modelling.

Some conclusions are that biological erosion processes are very important as key- processes. Positive feedback mechanisms are important a t various stages in the evolution of the patterns studied play an important role in creating

*iandccape and Environmental Research Çroup, University of Amsterdam. Nieuwe Primgracht 130, 1018 VL PAR.I. ,,, Amsterdam, The Netherlandc gpq?; k&p@ ,, ~

$:,>*,. P."-\ f& .- * $:?,.',:< $%p:

RDS AN IMPROVED LAND MANAGEMENT IN THE ARlD AND SEMI-ARID ZONES

micro- environments where vegetation recovers first following disturbances. This is particularly the case following wildfire. The fint post-fire rain produces patterns in ash and litter around sites where concentrating these at locations where shrubby vegetation subsequently resprouts or becomes seeded. On abandoned land, patterns evolve in time and reflect parent material grazing and the climate. Understanding how patterns at different hierarchical levels are related is a current research objective.

PART IV

TOWARDS AN IMPROVED LAND MANAGEMENT IN ME ARID AND SEMI-ARID

n the spatial pattern of herbqge oodland in Eastern Australia.

ER and Robert PALMER*

production

"hl * .--Y$

*:* INTRODUCTION $Y*='*='$? ?*gi$j 9 -.3@%3i( Y

hat is the productive significance of vegetation banding $-:$ seen in many arid environments? Are they simply - 9$ interesting patterns from the air, or is banding a visual 3qb& -

L"QF%g 9

expression of landscape pmesses? Banded vegetation patterns have <$;%yj been noted in many arid landscapes in Australia (Mabbutt G Rnning .mk : ,= !$ 1987, Ludwig G Tongway 1995) How do these vegetation bands * t w i $ y;- a. e change over time? What are the effects of grazing on banding? How .;y,$e does banding influence animal production? These questions were bqK$r

2 ,424, -̂ a%g;

addressed in a 7 year graring study in a banded mulga (Acacia aneura) ag.9 woodland in eastern Australia. , Y@ ~2

*:* METHODS

Within a mulga woodland in semi-arid eastern Australia. 200 ha was fenced into 12 plots varying in size from 7.5 ha to 30 ha. Beginning in 1986. six young merino sheep were annually placed into each of 6 plots and 6 sheep and 6 kangaroos were placed in the other plots. This created 6 stocking rates varying from 0.3-0.8 sheeplha. The design of "

the study is described by Wilson (1991) and the landscape is characterised by Tongway and Ludwig ( 1 990). Herbage (grasses and ' dicots), yield (standing biomass) and composition were estimated every three months from 100 fixed I m2 quadrats in each plot. Approximately 40% of the quadrats in each plot were in the runoff slopes (< 0.5%) spanely covered by the perennial grass Eragrostis eriopoda. 24% of quadrats were in the grass bands (toe of runoff slopes) densely covered

* CSlRO Division of Wildlife and Ecology. PO Box 84, Lyneham. ACT; Australia PART IV

DS AN IMPROVED LAND MANAGEMENT IN ME ARlD AND SEMI-ARID ZONES

by perennial grasses (e.g. Monachather paradoxa and Thyridolepsis mitchelliana), 29% of quadrats were in bands of mulga trees (groves) which had an undentorey of other grasses (e.g. Digitaria spp) or litter, and 4% of the quadrats were classified as 'other' (e.g. log and termite mounds) and were not included in the analyses. The effects of stocking rate. grazing treatment (sheep vs sheep + kangaroos). landscape zone and time and their interactions on herbage yield. were tested by analysis of variance of repeated measures (Censtat 1993).

*:* RESULTS

Herbage yield was significantly greater (p > 0.001) in the two runon zones (grass and mulga bands) and varied through time due to rainfall and consumption. In those plots lightly and moderately grazed (c0.6 sheeplha), herbage yield was on average 8 times greater in the runon grass and mulga bands compared to the open runoff slopes, but in some cases was over 30 times greater in the runon bands. The herbage yields in the grass bands were generally 2-7 times greater than in the mulga bands. (e.g. Fig. la). In response to small rainfall events, herbage was produced in the gras bands (zones immediately upslope of the mulga trees); little was produced in the mulga bands and runoff slopes (e.g. 1993. Fig. la). In the heavily grazed plots (> 0.6 sheeplha). the yield of herbage was lower and uniform across al1 three landscape bands (e.g. Fig. I b). Sheep production collapsed in those plots where herbage banding was destroyed.

Craph

Figure 1. The temporal and spatial pattern of herbage yield (standing biomass) in a mulga woodland in eastern Australia. (a) a lightly grazed plot (0.3 sheeplha) and (b) a heavily grazed plot (0.8 sheeplha).

*:* DISCUSSION

In the semi-arid woodland described here. the pattern of herbage production is highly patterned and persistent for yean under light to moderate grazing. As predicted by the landscape analysis of Tongway and Ludwig ( 1 990). perennial herbage production is far greater in runon landscape zones (grass and mulga bands) which receive nutrients and water from the runoff slopes. In some cases, the runon grass bands

PART IV

TOWARDS IMPROVED LAND MANAGEMENT IN ME ARlD AND SEMI-ARID

(less than 25% of the landscape) produced > 95% of the herbage.

These landscapes are patterned at a range of scales from mulga tree band-interband to grass tussock-intertussock (Ludwig G Tongway 1995). The fine scale pattern of banded herbage production can be quickly destroyed through over-grazing - tiger bush can loose its stripes This destruction results in an inability of these landscapes to capture local runoff of water and rich sediments. The loss of patchy resource conservation is a major cause of landscape dysfunction or degradation as defined by Tongway and Ludwig (1 994). When herbage banding is destroyed by over-grazing, the landscape fails to respond to most rainfall events - animal production collapses. When tiger bush looses i stripes, it no longer roan with production.

*:* R E F E R E M C E S

CENSTAT (1 993) Release 5.3 1 , Lawes Ag. Trust (Rothamsted Expt. Station).

LUDWIC and TONCWAY (1995). Land Ecol. 10. 51-63.

MABBUTT and FANNINC (1 987). J.Arid Environ. 12, 41-59.

TONCWAY and LUDWIC (1990). Aust. J. Ecology. 15, 23-34.

TONCWAY and LLlDWlC (1994). Pac. Cons. Biol; 1, 201-08.

WlUON (1991). Range]; 13. 69-80.

PART IV

er dams: similarities and differences

REYS, D.M. HART, P.B. MITCHELL

chool of Earth Sciences, Macquarie University, Australia 2109.

Vegetation arcs are near parallel bands (spacing 25-300 m) aligned approximately perpendicular to slope.

widely reported from arid and semi-arid areas on low angleslopes in Africa and Australia.They support distinct communities of grasses, shrubs and trees. Heavy episodic tains result in sheetwash which transports Sand bedload, clay suspended load and floating load of leaves. charcoal. seeds and faecal material, but importantly rilling does not occur.

Litter dams are similar to vegetation arcs in shape and orientation but they are considerably smaller (spacing O. 1-2 m). They are common in sub-humid to humid bushfire prone south-east Australia on low angle hillslopes subject to rainwash along anastomosing drainage lines. The convex downslope dams are composed of floating load material. ash, fragments of charcoal. and linear organic debris. Downslope of the dam is a steeper scarp or erosion zone. Trailing edges of Iitter dams often merge with othen forming an interconnected network over the hillslope i e. in a similar way to vegetation arcs. Upslope of the dam a microterrace forms as bedload mineral material, mainly quartz sand- sized grains, is deposited in the ponded zone created by the litter dam. Within the microterrace a variety of crusts form during the post-fire period as the microterrace is stabilised by algal and root mats but is largely devoid of higher plants. In this sense the bare microterrace resembles the vegetation deficient intergrove of the vegetation arcs. The litter dam and scarp are the sites of regeneration of fire tolerant species and stoloniferous plants, rushes and sedges. seedlings of herbs and

PART V School of Earth Sciences. Macquarie University, Australia 2109

perennial grasses and fire dependent obligate seeding shrubs. The pattern of these microrelief forms has been shown to persist for longer than a decade between major fires. Significantly, in areas patterned with litter dams and microterraces rilling does not occur

Data(n > 70) extracted from the literature and from Our own field measurements show that a strong inverse association exists between wavelength (i.e. spacing between arcs or dams) and hillslope gradient, but the data are grouped in two distinct subsets:

(i) vegetation arcs with slopes gradient 0.002- 001 mlm and wavelength 25-300 m

(ii) litter dams with slope gradient 0.01 5-0.1 and wavelength 0.1-2 m.

There is a noticeable gap spanning wavelengths of 2- 25m and slope gradients between 0.01-0.01 5 rnlm despite the fact that similar transport processes occur across this range. This gap raises some interesting questions. Are the two subsets clearly separate or are they part of a continuum? Does the data reflect the scale or manner of initial recognition and study. i.e. vegetation arcs on a regional scale from the air and litter dam on a local scale from ground level? What influence does rainfall have on the development of arcs and what role does plant life-form play? Do litter dams occur as a subset within the larger scale patterns?

We report the results of a preliminary investigation in mallee (open woodland of multistemmed Eucalyptus viridis) within the sub-humid Pilliga State Forests. This shows spatial organisation at an intermediate scale (average wavelength 12 m on a gradient 0.01 5mlm) and suggests that the relationship between wavelength and gradient holds true on slopes between about 0.002 and 0.2 mlm.

Within the same mallee community litter dams occur on locally steeper slopes (0.0 6mlm). This suggests that in some sites litter dams with associated crusts forma subset of larger vegetation patterns.

PART V ,~~~ ,.<, g;&rp@ g$&~*~ ,,,, ;-...".>c:- 2 :*>.;*;$$<*

ruinfull, soil-uater, and herbage production nes in semi-arid wooded rangelands

he response in soil-water (0-300 mm) and herbage production in three contiguous geornorphic zones in a serni-arid rnulga woodland was measured after a series of cool- and warrn-season rainfall events.

The soil-water rneasurernents showed that for each of the rainfall events. pronounced redistribution of rainfall from runoff (source) into runon (sink) zones occurred, redistribution increasing with the arnoun of rainfall. The arnount of soil-water in the 0-300 mm layer followed the order: rnulga grove > interception zone > runoff zone.

Sirnilarly, redistribution of rainfall resulted in significant differences in the arnount of herbage production in each of the three zones and followed the order: rnulga grove > interception zone > runoff zone.

lrnpairrnent of surface hydrology following the erection of rnetal barriers around plots in both source and sink zones had a significant effect on dry rnatter production by resident perennial gras populations. In t runoff zone. production by woollybutt (Eragrostis eriopoda) was significantly enhanced when incidnet rainfall was contained in situ

Conversely, dry rnatter production by rnulga grass (Thyridolepis rnitchelliana) in the lower interception was significantly depressed by barriers preventing access by overland flow.

The use of these data in the dweloprnent of a herbage production rnodel is discussed.

CSlRO Diu. Wildlife and Ecology, PO Box 84 Lyneham ACT 2602 PART V

U ndex des

ABBADIE L., 73

ALLOCIA M., 79

BELI-IER C., 31

BERCKAMP C., 33

BlANClOrrO O., 79

BROMLEY 1.. 49

BROUWER]., 49

BROWN K., 41

CAMMERAAT L.H.. 87

CERDA A.. 33

CHAPPEL A.. 75

CHARLTON M., 75

COUTERON P., 63

D'HERBES J.M., 25, 37. 75. 77

DEL BARRIO C., 79

DUNKERLEY D., 41

EDDY J., 95

EDMUNDS W.M., 49

EHRMANN M., 27.45

. FANNINC P.C.. 95

FELLMAN E.. 49

FREUDENBERCER D., 89

CALLART F., 79

GALLE S., 27, 45. 65

GAZE S.R., 49

GREENE R.S.B.. 97

GUILLAUME K., 73

HAMANI S., 25

HART D.M., 95

HIERNAUX P., 83

HUMPHREYS G.S., 95

ICHAOU A., 77

IMESON A., 33, 87

JANEAU J.L., 35

JONGEJANS].. 33

JUO A., 53

KAPPAS M., 47

LEFEVER R.. 71

LEJEUNE O., 71 LEPAGE M., 59

LLIDWIG 1.. 67

MACDONALD B.C.T. MALAM 1.. 31

MANU A., 53

MARSDEN S., 67

MAUCHAMP A., 35 MELVILLE M.D.. 39

MITCHELL P.B.. 95

MOUGENOT B.. 25.

NOBLE J.C.. 57. 97

OUEDRAOGO P., 59

PALMER R., 89

PEUGEOT C.. 45

PUICDEFABREGAS 1.. 79

RAJOT J.L., 3 i

ROSSElTl Ch.. I l

SEGHIERI J., 27, 65 SUDLOW J., 49

TARIN G., 35

TAUPIN J.D., 49

THUROW TL., 53

TRICHET J., 31

VALENTIN C.. 27. 31, 37, 75

WARREN A., 75

WHITE 1.. 39

ZANGUINA l., 53

ZONNEVELD I.., 21