<J \

ISSN 0366-2047

CèAàZ. À)U

BOLLETTINO DELLA SOCIETÀ DEI NATURALISTI IN NAPOLI

VOLUME CII - 1994/1995

"SbVAVTHS QNbqfc

4UG 0 4 1997

GIANNINI EDITORE NAPOLI 1996

ISSN 0366-2047

BOLLETTINO DELLA SOCIETÀ DEI NATURALISTI IN NAPOLI

VOLUME GII - 1994/1995

GIANNINI EDITORE NAPOLI 1996

SOCIETÀ DEI NATURALISTI IN NAPOLI

VIA MEZZOCANNONE, 8

CONSIGLIO DIRETTIVO 1994/96

Prof. Pietro Battaglini	- Presidente
Prof. Enrico Franco	- Vice-Presidente
Prof. Teresa de Cunzo	- Segretario
Prof. Maria Carmela Del Re	- Vice-Segretario
Prof. Eugenio Piscopo	- Tesoriere
Prof. Amalia Tavernier	- Bibliotecario
Prof. Maria Rosaria Ghiara	— Redattore delle Pubblicazioni
Prof. Giuseppe Caputo	- Consigliere
Prof. Oreste Schettino	- Consigliere
Prof. Luciano Ferrara	- Consigliere
Prof. Mario Milone	- Consigliere

REVISORI DEI CONTI

Prof. Silvio Di Nocera Prof. Fimiani Pellegrino

REVISORI DEI CONTI 1996/97

Prof. Sergio Scippacercola

Dr. Gabriele De Filippo

Prof. Silvio Di Nocera (supplente)

BOLLETTINO DELLA SOCIETÀ DEI NATURALISTI IN NAPOLI

Comitato redazionale

Claudio Agnisola Filippo Barattolo Pietro Battaglini Giuseppe Caputo Graziano Fiorito Enrico Franco Giovanni Fulvio Russo

Maria Rosaria Ghiara - Redattore delle pubblicazioni

INDICE

Russo D. - Preliminary observations on thè behaviour displayed at thè feeding stations in a captive colony of Rodrigues fruit bats Ptero¬ pus rodricencis (Mammalia: Chiroptera) . . . pag. 5

Capaldo A., Laforgia V., Varano L., Cavagnuolo A. - The adrenal gland of some lizards living in xeric habitats: morphology and NA/A celi ratio . » 15

Palumbo F., Rapolla E. - Mean sea level and seismic warning . » 27

Palumbo F. , Rapolla E. - Mean sea level fluctuations and volcanic activity » 35

Barbera C., Lamagna R., Signore M. - Triassic estherids from Salerno

(Southern Italy): preliminary notes . . » 43

Barbera C., Bilancio F., Giordano P. - A miocenic reef near Catanzaro,

Calabria. South Italy . » 49

Signore M. - Deinonychosauria (Saurischia, Theropoda): a brief group

synopsis . . » 57

Pugliano M.L., Murolq M., Buondonno C. - Study on Glass components from Soils of Phlegrean Fields. First Note: The glasses in thè sands of Monte Spina (Agnano) . . » 73

Fulgione D., Mirabella P., Milone M. - Analytical procedure to study

animai communities structure's in fragmented environments .. » 81

Fulgione D., Rusch C.E., Milone M. - Fire in thè mediterranean Scrub:

Evaluation of thè Impact in thè Licola-Castelvolturno Reserve

(Ce) Through thè Birds . » 91

CONTRIBUTI IN ITALIANO

Scippacercola S., Amenta P., D’Ambra L. - Il Partial Least Squares e l’Analisi in Componenti Principali rispetto ad un sottospazio di riferimento. Un’applicazione per la rilevazione di inquinamento di ambienti fluviali . . » 101

Bruno P.P.G. - Esempi di applicazione del metodo sismico a riflessione per indagini superficiali in alcune aree campioni dell’Appennino Campano-Lucano . » 117

4

Barbera C. - Le collezioni Paleontologiche del Museo di Paleontologia dell'Università di Napoli «Federico II». I Vertebrati pleistocenici.

1, I Mammiferi di Venosa . . . . . . pag. 137

Soppelsa O., Sagnibene P., Battaglimi F. - Considerazioni ecologiche

sugli Eterotteri del Fiume Lete (Campania) . » 155

Soppelsa O., Sagnibene P. - Nuovi rinvenimenti di Eterotteri nell’area

settentrionale della Provincia di Caserta (Campania) . » 169

CONFERENZE E COMMEMORAZIONI

Casertano L. - Attività Vulcano Poas: Meccanismo unico delle sue

diverse manifestazioni . . . . . » 177

Ariani A. - Ricordo di Mario Salfi nel 25° anniversario della sua

scomparsa . . . . » 207

Russo L.F. - Ricordo di uno scienziato: Filippo Silvestri . » 223

Boll Soc . Natur . Napoli - Voi 102 ( 1994-1995 ): 5-14

Preliminary observations on thè behaviour displayed at thè feeding stations in a captive colony of Rodrigues fruit bats Pteropus rodricensis (Mammalia: Chiroptera)

Danilo Russo

Key words: bat; behaviour; captive breeding; feeding activity; Pteropus ; territoriality; zoo.

Abstract. Feeding activity and other behaviours displayed at thè feeding sites were studied in a captive colony of Rodrigues fruit bats, Pteropus rodricensis , at thè Jersey Wildlife Preservation Trust, Jersey (U.K.).

All thè behaviours occurring at thè feeding stations were recorded.

The temporal change in thè number of bats at thè feeding stations was also recorded during thè 2 hours following thè time food was presented.

The results showed that some males defended well defined territories that included a feeding site and that were habitually frequented by several females.

The colony showed a polygamous resource-based mating System, as thè terri- torial males could copulate with all thè receptive females entering their respective territories.

Moreover, one of thè feeding sites was used and defended by females.

Some of thè non-territorial bats (probably low-ranked subjects) moved on thè ground searching for food.

Aggressive behaviour displayed by roosting bats towards individuata moving on thè ground should indicate that territories could also comprise floor areas.

Riassunto. Osservazioni preliminari sul comportamento esibito in catti¬ vità alle stazioni di alimentazione in una colonia di pipistrelli frugivori di Rodrigues Pteropus rodricensis (Mammalia: Chiroptera).

L’attività alimentare ed altri comportamenti manifestati presso i siti di alimen¬ tazione sono stati studiati in una colonia di pipistrelli frugivori di Rodrigues, Pteropus rodricensis, in cattività presso il Jersey Wildlife Preservation Trust, Jer¬ sey (U.K.).

Received 8.10.94, accepted 30.1.95

6 Boll Soc . Natur. Napoli - Voi 102 ( 1994-1995 )

È stata misurata la variazione temporale del numero di individui presso le stazioni di alimentazione durante le due ore successive alla presentazione del cibo agli animali, registrandone nel contempo i comportamenti osservati.

I risultati hanno mostrato che alcuni maschi occupavano territori ben definiti includenti un sito di alimentazione ordinariamente frequentato da più femmine.

I maschi territoriali potevano accoppiarsi con tutte le femmine recettive penetrate nel loro territorio, risultandone un sistema di poligamia con difesa delle risorse.

Uno dei siti di alimentazione veniva utilizzato quasi esclusivamente, nonché difeso, da femmine.

Alcuni soggetti (probabilmente di basso rango sociale) erano soliti alimentarsi cercando cibo a terra.

Aggressioni compiute da pipistrelli occupanti rami posti a piccola altezza dal suolo dirette ad individui in movimento su questo hanno portato a proporre un'ipotesi di territorialità estesa ad aree del pavimento.

Introduction

The Rodrigues fruit bat, Pteropus rodricensis Dobson 1878, is a very endangered species from thè order Chiroptera, suborder Megachiroptera.

This species is extinct in Mauritius (Mascarene Islands), where it is known from subfossil material (Cheke & Dahl, 1981); today it is found only in Rodrigues, a small volcanic island located about 650 km east of Mauri¬ tius.

The overgrowing human population is heavily damaging thè fragile insular environment by clearing thè forests, thè typical habitat of thè species, to use those areas for grazing and agriculture (West & Redshaw, 1987).

Deforestation affects thè bats in two ways: firstly, reducing feeding and roosting areas, and secondly, making thè bats more susceptible to thè effects of cyclones that periodically pass over Rodrigues (West and Red¬ shaw, 1987).

Since 1976 thè Jersey Wildlife Preservation Trust (Jersey, U.K.) has been running a captive breeding programme for this species (Durrell, 1976).

In 1994 thè Jersey Zoo holded 73 specimens, living in two different enclosures, one of which was on exhibit.

The programme is successful and represents one of thè main hopes for thè survival of P. rodricensis.

Russo D. - Behaviour of captive Rodrigues fruit bat 7

Behavioural studies on captive populations can contribute to improve their management, especially for those species which display a complex social behaviour, as in thè case of P. rodricensis (Carroll & Mace, 1988).

The present study deals with thè feeding activity and other behaviours displayed at thè feeding sites in a captive colony of Rodrigues fruit bats. The data were collected at thè Jersey Zoo during 1994 Breeding and Conservation of Endangered Species Summer School.

Materials and methods

At thè time thè data were collected thè enclosure where thè bats were housed was rectangular-shaped, measuring about 7x5x4 m high.

Several wood perches, from which plastic dishes containing food were suspended, provided feeding and roosting sites.

Some other food and water containers were suspended from thè enclosure walls.

Lighting consisted of a reversed day-night cycle with a two hours annual variance, equal to thè one of Rodrigues.

Night time was simulated using a soff light of twilight intensity; for daylight, a light with a spectrum very similar to sun light was provided, together with an UV lamp to allow a naturai synthesis of vitamin D.

Three thermostatically controlled electric heaters permitted a fixed “nocturnal ” temperature of 21° C, and a “ diurnal” temperature of 26-28° C (Carroll, 1978). Further details about husbandry are given in Carroll (1978).

The data were collected from thè 6th to thè 13th of August 1994.

During thè first 8 hours of data collection, an '‘ad libitum” sampling method (Martin & Bateson, 1986) was used in order to learn to recognize individuals and thè areas they used and to identify thè behavioural reper- toire of thè species.

A second phase of 12 hours, divided in 6 sessions of 2 hours, was carried out using a 1-0 sampling method (Martin & Bateson, 1986); each scan interval lasted 30 sec. Each session of data collection started at about 9 a.m., when food was provided to thè bats.

Most of thè bats, except thè young ones, were marked with coloured rings on thè thumb to facilitate their identification.

Territorial subjects (see Results) were named with thè initial letters of thè colour of their rings: S (silver), R (red), YB (yellow-blue), RB (red-blue), OR (orange-red).

8

Boll Soc. Natur. Napoli ~ Voi 102 (1994-1995)

Sex of adult bats was quite easily identified, as males had well devel- oped genitalia and were tagged on thè left thumb, while females were ringed on thè right.

During data collection, sex of immature individuals was not recorded as it was difficult to establish at a distance, and as they were not marked.

At thè end of each scan interval of 1-0 sampling phase - marked by a beep-sounder - number, age class (juvenile-adult), sex (of adults) and activity of thè subjects observed at thè stations during thè previous 30 sec were recorded.

Time-sampling observations were focused on 4 feeding sites: 3 of them were plastic dishes hung on branches, and one was represented by thè floor, where bats could feed on fruit fallen front thè suspended bowls. They were observed in turn in a clock-wise cycle, each one for thè dura- tion of a scan interval. These sites were chosen as far as possible from one another so as to avoid influences among individuals at different sites.

During “ad libitum” sampling thè four sites were simultaneously checked together with other 3 bowls.

The behaviours recorded during 1-0 sampling are described in thè following 7 categories:

non-interactive behaviours

a) Feeding. All thè behaviours directly involving thè use of food, such as searching through thè food for some favourite items, holding and eating food etc.

b) Roosting. Carroll (1979) distinguished between two resting pos- tures, one (actual roosting) occurring when thè animai is asleep, thè other one (hanging alert) occurring when thè animai is awake but stationary.

In thè present study, both thè postures were recorded as “roosting”.

c) Self-grooming. Licking, nibbling fur or scratching body with one hindfoot.

d) Urination - Defecation. interactive behaviours

a) Aggressive behaviour. Any agonistic action, such as stealing food, chasing an opponent and performing “wing shake”, an aggressive action which is undertaken outstretching one or both wings in front of thè body (Carroll, 1979).

b) Mutual contact. This category comprises any non-aggressive, non- sexual physical contact between two subjects, such as social grooming.

c) Sexual behaviour. Any behaviour displayed during sexual activity:

Russo D. - Behaviour of captive Rodrigues fruit hai

9

male-female grooming, male contacting female's genitalia, actual copula- tion and post-coitum mutuai interactions.

The data on temperai trends and on activities at sites 1 and 3, where a territorial male was found (see Results), were pooled together and expressed as mean values.

The significance of differences among means was tested using 95% confidence limits, and a chi-square test was applied to evaluate whether thè observed differences in thè number of scans during which a category of individuate (territorial male; non-territorial male; female; juvenile) was observed at each site could have been due to chance fluctuations.

The correlation between number of individuate on thè ground and time was tested for significance with thè Spearman rank-order correlation coefficient rs and thè combined t statistic (Siegei & Castellan, 1988).

Results

“Ad libitum ” sampling observations

Many males, and all females and juveniles, ranged within thè whole enclosure and used any feeding station available, including thè floor.

Five males (S, R, YB, RB, OR) , however, tended to stay at a particular feeding site which was very rarely left.

In fact, 100% of thè recorded positions occupied by these males during thè entire “ad libitum" observation period resulted comprised in an about 3 metres-diametered circumference centred on thè bowl they used.

In thè case of S male, this area appeared reduced to a narrower, semicircular surface bordered by two of thè enclosure walls.

These males often rubbed their neck and chest on thè branches and other objects of thè area they occupied and on any approaching female (scent-marking behaviour).

When a female was marked in this way, thè male could inspect her genitalia; copulation was sometimes observed after this interaction.

Resident males showed an active territorial behaviour, as they ah lowed females to enter thè area they used, but all thè other bats which tried to reach thè bowl - both adult males and juveniles - were chased .

Territorial behaviour was not shown by individuate on thè floor.

On 16 occasions, bats roosting on low branches attacked nearby subjects which were crawling along thè floor , flying and skimming over them.

10

Boli Soc . Natur. Napoli - Voi. 102 ( 1994-1995 )

Time - sampling observations

Site 2 was used by a higher number of animals (N = 4.27 ± 0.05, mean ± 95% C.L.) than thè others (N site 1 = 1.83 ± 0.15; N site 3 = 0.99 ± 0.07; N floor = 1.00 ± 0.1).

The presence of a territorial male was observed at sites 1 (“R” subject) and 3 (“ RB” subject). The low number of bats seen at these sites was due to thè active defence performed by territorial males.

Tab. 1 shows thè number of scan intervals during which thè presence of a territorial male, a non-territorial male, a female, and a juvenile was recorded. This number was significantly higher for territorial males at thè site they occupied, in comparison with non-territorial males and juveniles (site 1: x2 = 281.4, di. = 2, p <0.001; site 3: x2 - 189.2, di. = 2, p <0.001). Females were normally accepted.

Table 1 - Number of scan intervals during which thè presence of a territorial male (T. male), a non-territorial male (N.t. male), a female and a juvenile (Juv.) was recorded. Total number of scan/site = 360.

Site	T. male	N.t. male	Female	Juv.
1	229	45	318	14
2	-	36	338	23
3	213	7	117	11
Floor	-	138	84	84

Site 2 was constantly occupied by one or more females; on several occasions, they were seen chasing juveniles and males which attempted to feed. For this reason thè number of intervals during which a female was recorded at site 2 was strikingly higher in comparison with these other two categories (x2 = 608.2, di. = 2, p < 0.001). Site 2 was thè only feeding station frequented by females holding a newborn.

No territorial males were seen crawling on thè floor to search for food.

Fig. 1 shows thè average temporal variation in thè number of individ- uals during thè 2 hours following food presentation.

At sites 1 and 3, this number oscillated around a value dose to 1 (N = 1.41), again due to thè Constant presence of thè territorial male.

Wider oscillations of this number were recorded at station 2, where it decreased after about 94 minutes.

Russo D, - Behaviour of captive Rodrigues fruii bat

11

During thè first 4 minutes no bats were seen on thè ground, then thè number of subjects tended to increase with a highly significant positive correlation (rs = 0.98; t = 37.5; d.f. = 58; p 0.001).

scan

Figure 1 - Mean temporal variation in thè number of individuate during thè 2 hours following food presentation. Vertical bars show thè standard error of mean.

The percentage of scan intervals during which each behavioural cat- egory was recorded is shown in fig. 2. Feeding stations 1, 2 and 3 were extensively used as roosting sites.

Mutual contact was not frequently recorded.

Sexual interactions observed at sites 1 and 3 all occurred between thè territorial male and thè females entering thè territory.

Aggressive behaviours were recorded in two main situations: when a territorial violation occurred, and when two or more bats tried to obtain priority over food.

Feeding was thè only frequent behavioural category shown by thè bats on thè floor. These individuate were often observed picking up some food with their mouth and then reaching a perch where food was ingested.

The aggressive events observed on thè floor (6.6 % of thè total activity) were triggered by competition for food.

12

Boll. Soc. Natur. Napoli - Voi. 102 (1994-1995)

Sites 1 and 3 (tnean)

□ Aggressive behaviour ■ Mutual contact

0 Sexual behaviour 0 Feeding

□ Roosting H Grooming

□ Urination - defecation

Floor

Figure 2 - Percentage of scan intervals during which each activity was recorded at sites 1, 3 (a), 2 (b) and floor (c). Total number of time intervals = 360.

Discussici

The results indicate that thè captive colony of P. rodricensis used a polygamous trophic resource-based mating System: high-ranked males defended a territory where food was concentrated, and mated with all thè females in oestrus which frequented thè area.

This polygamous System is known for several species of vertebrates: among mammals such as thè impala (Aepyceros melampus) and thè vicuna ( Lama vicugna ) and also among birds (Alcock, 1989).

Feeding territories also resulted to be defended in a colony of P. rodricen¬ sis enterily composed of males at Poznan' Zoo, Poland (Russo, pers. obs.).

13

Russo D. - Behaviour of captive Rodrigues fruii hat

, ■ i::-' " ' ... ”v ’ 7% Wy I" W W

Few studies have been carried out on wild Rodrigues fruii bats (Dur- rell, 1976; Carroll, 1981) and il is unknown whether thè species uses such a strategy in naturai conditions, where food resources are distributed in a very different way in comparison with captive conditions.

During thè present study, mostly carried out during nocturnal condi¬ tions (i.e. when bats were more active), territorial males roosted at thè feeding sites they defended.

Carroll (1979; 1988) and Carroll & Mace (1988) claimed that in cap¬ tive colonies of P. rodricensis , males defend both roosting areas and feeding territories. Females associate with different males during feeding periods than while roosting, and may mate with more than one male within minutes, merely by Crossing from one territory to another (Carroll & Mace, 1988).

Defence of feeding sites by females, as observed at station 2 in thè present study, was not reported in previous researches.

Floor feeding, already observed in captive Rodrigues fruit bats (Car¬ roll, 1979; Carroll, 1988) seems to be performed by low-ranked individuals and, among these, by all those adult males which do not yet own a territory.

This behaviour is very unlikely to be performed in thè wild; in fact, it would be very disadvantageous, because crawling on thè ground would involve a high energetic expense and a great risk of predation.

A reasonable hypotesis is that floor feeding is induced by captivity in a situation of enclosure overcrowding, where trophic territories are avail- able for only few bats.

The initial period of scan sessions when no bats were observed on thè ground may be related to thè absence of food, as it had not yet fallen from thè suspended bowls; then thè number of individuals progressively aug- mented, probably as thè quantity of food on thè floor increased.

Attacks performed by bats from low branches to approaching subjects on thè floor could indicate that territories also comprised ground areas, yet this phenomenon should deserve further investigation.

Acknowledgements

I am grateful to thè International Training Centre personnel of thè Jersey Wildlife Preservation Trust and particularly to Dr. John Fa and Mr. Chris Clark.

14

Boll. Soc. Natur. Napoli - Voi 102 (1994-1995)

Thanks to Dr. David Houston, Edinburgh University, and to Dr. Richard Griffiths, Durrell Institute of Ecology (University of Kent), for all thè precious suggestions about thè methods of data collection and analy- sis.

REFERENCES

Alcock, J., 1989. Animai behaviour, 4th ed., Sinauer Associates, Ine.

Carroll, J.B., 1978. Behavioural observations on thè Rodrigues fruit bat following a move to new accomodation and reversai of light/dark regime. Dodo, J. Jersey Wildlife Preservation Trust, 15: 52-60.

Carroll, J.B., 1979. The generai behavioural repertoire of thè Rodrigues fruit bat Pteropus rodricensis at thè Jersey Wildlife Preservation Trust. Dodo, J. Jersey Wildlife Preservation Trust, 16: 51-59.

Carroll, J.B., 1981. The wild status and behaviour of thè Rodrigues fruit bat Pteropus rodricensis. A report of thè 1981 fìeld study. Dodo, J. Jersey Wildlife Preservation Trust, 18: 20-29.

Carroll, J.B., 1988. The conservation programme for thè Rodrigues fruit bat Pteropus rodricensis. In B.L. Dresser, R.W. Reese, E.J. Maruska (eds.). Proc. Vth Conference on Breeding Endangered Species in Captivity , Cincinnati : 457-475.

Carroll, B. & Mace, G., 1988. Population management of thè Rodrigues fruit bat. Int. Zoo Yb., 27: 70-78.

Cheke, A.S. & Dahl, J.F., 1981. The status of bats on western Indian Ocean islands, with special reference to Pteropus. Mammalia, 45 (2): 205-238.

Durrell, G., 1976. The Mauritian expedition, Jersey Wildlife Preservation Trust Annual Report. 13: 7-11.

Martin, P. & Bateson, P., 1986. Measuring behaviour: an introductory guide, Cambridge University Press.

Siegel, S. & Castellan, N.J., 1988. Nonparametric statistics for thè behavioural Sciences, 2nd edition, McGraw-Hill International Editions.

West, C.C. & Redshaw, M.E., 1987. Maternal behaviour in thè Rodrigues fruit bat Pteropus rodricensis. Dodo, J. Jersey Wildlife Preservation Trust, 24: 68-81.

Boll Soc . Notar . A'apo/f - Lo/. 702 (1994-1995): 15-25 15

The adrenal gland of some lizards living in xeric habitats: morphology and NA/A celi ratio

Capaldo Anna, Laforgia Vincenza, Varano Lorenzo, Cavagnuolo Antimo

Dipartimento di Biologia Evolutiva e Comparata, Università degli Studi Federico II via Mezzocannone 8, 80134 Napoli, Italy

Key words: Adrenal gland morphology of some lizards.

Abstract. The Authors have studied thè adrenal morphology, thè ratios between noradrenaline (NA) and adrenaline (A) cells, and their relationships with phylogenetic position in six species of thè family Lacertidae: Mesalina olivieri , Eremias strauchi, Eremias pleskei, Acanthodactylus pardalis, Acanthodactylus bo- skyanus, Latastia longic andata.

The adrenal gland of all thè species appears elongated and surrounded by a thin connective envelope. The chromaffin tissue is concentrated in M. olivieri at thè head pole, in a cap, and in a thin, discontinuous dorsal ribbon, with few islets scattered in thè steroidogenic parenchyma. In thè other species there is a higher degree of admixing of thè two types of tissue, with many islets interspersed in thè interrenal tissue.

In all thè species thè NA cells are located in thè outer layers of thè superficial chromaffin tissue, while A cells occupy thè inner layers and are thè only consti- tuents of thè chromaffin islets.

On thè basis of thè different values of thè NA/A celi ratio, we suggest that M. olivieri, with a high ratio (3.1/1) should be considered an ancient species, while thè other species, with a ratio around 1 / 1 , should be considered of more recent origin.

Riassunto. La ghiandola interrenale di alcuni Lacertidi diffusi in am¬ bienti aridi: morfologia e rapporto cellule NA/A.

Gli Autori hanno studiato la morfologia della ghiandola interrenale, i rapporti tra cellule a noradrenalina (NA) e adrenalina (A), e le loro relazioni con la posizione filogenetica in sei specie di Lacertidi: Mesalina olivieri , Eremias strau¬ chi, Eremias pleskei , Acanthodactylus pardalis, Acanthodactylus boskyanus , Lata¬ stia longic andata.

La ghiandola appare in tutte le specie allungata, circondata da una sottile capsula connettivale. Il tessuto cromaffine é concentrato in M. olivieri al polo

Received 16.12.94, accepted 30.1.95

Boll Soc . Natur. Napoli - Voi 102 ( 1994-1995 )

cefalico, in un cappuccio, e in un cordone dorsale sottile e discontinuo, con pochi isolotti sparsi nel parenchima steroidogenetico. Nelle altre specie c’è un maggior grado di contatto tra i due tipi di tessuto e un maggior numero di isolotti sparsi nel parenchima.

In tutte le specie le cellule NA occupano gli strati esterni del tessuto crom af¬ fine di superficie, mentre le cellule A sono presenti negli strati interni e sono le sole costituenti degli isolotti.

In base ai valori del rapporto cellule NA/A, suggeriamo che M. olivieri, con un rapporto alto (3.1/1), possa essere considerata una specie conservativa, mentre le altre specie, con un rapporto vicino a 1/1, possano essere considerate di origine più recente.

Introduction

Previous studies on thè morphology of thè adrenal gland of Squamata showed a great variability in thè topographic relationships between thè steroidogenic and thè chromaffin tissues, in thè distribution of thè two types of chromaffin cells and in thè ratio between NA cells and A cells.

Early authors attributed a very uniform arrangement to this gland in Squamata, describing a steroidogenic parenchyma, containing small islets formed by adrenaline (A) cells, and a continuous dorsal ribbon formed by noradrenaline (NA) cells, sending digitations into thè steroidogenic paren¬ chyma (Wright & Chester Jones, 1955, 1957; Chester Jones, 1957 a,b; Gabe & Martoja, 1961; Wasserman & Tramezzani, 1961, 1963; Gabe, 1970).

Nevertheless studies on thè adrenal gland of many species of Squamata did not confimi this generai uniformity, but showed many differences in thè arrangement of this gland (Varano et al, 1969; Saint Girons, 1976; Varano & Laforgia, 1976; Laforgia & Varano, 1978), also between species belonging to thè same family (Laforgia et al, 1982, 1983, 1985, 1990, 1993 a,b; Laforgia & Varano, 1982; Varano & Laforgia, 1982, 1991; Capai do et al, 1991) or to thè same genus (Laforgia et al, 1991).

In Podarcis s. sicula, e.g., thè NA cells are located in thè outer layers of thè chromaffin dorsal ribbon, while thè A cells are present in thè inner layers, in thè digitations and in thè chromaffin islets scattered in thè steroidogenic parenchyma.

Also P. wagleriana, P. muralis, P. peloponnesiaca and P. melisselensis show thè same pattern, while in P. hispanica and in P. taurica thè chromaf¬ fin and thè steroidogenic tissues are separated, and thè former is located in P. hispanica at thè head pole of thè gland, and at thè two poles in P. taurica.

| Càpaldo A | et al - The adrenal gland of lìzards

17

The study of thè comparative morphology of Squamata adrenal gland suggested a relationship between thè phylogenesis, thè morphology of thè gland and thè NA/A celi ratio: older species have more NA cells in thè adrenal gland» where thè steroidogenic and thè chromaffin tissues are more separated» while species of more recent origin have more A cells and show a high degree of admixing of thè two tissues.

So P. taurica and P. hìspanìca, with a high degree of separation between thè two tissues» and high NA/A celi ratios (3,6/1 and 2,6/1 respec- tively) can be considered more ancient species, while P. sìcula, P. wagle - riana, P, muralis, P. peloponnesiaca and P. melisselensis form a clearly homogeneous group of more recent origin.

Al so in birds where Ghosh (1977) demonstrated a connection between thè phylogenetic position of thè species and thè NA/A celi ratio, NA cells are more numerous in species of more primitive ancestry.

In this paper we have studi ed thè morphology of thè adrenal gland, thè distribution and thè relation ships between thè chromaffin and thè steroidogenic tissues» and between thè two types of chromaffin cells of specimens of six species belonging to thè famìly Lacertidae» in order to evaluate whether thè conclusions on thè phylogenetic position of thè species» as determined by adrenal morphology» are in agreement with thè results of other workers» applying biochemical» immunological or karyo- logical tecnichs to thè study of thè taxonomy within thè family Lacertidae.

Material and methods

The adrenal glands of thè following species were investigated: Mesa¬ lina olivierì, Eremias strauchì, Eremias pleskei » Acanthodactylus pardalis, Acanthodactylus boskyanus, Latastia longìc andata. The glands were fixed in a mix ture of potassium dichromate and sodium sulphate (buffered at ph 4.1 with acetato buffer» 5 M) to which 10% formaldehyde was added before use (Wood, 1963). The glands were embedded in parai fin, sectioned at 7 p and stained with one of thè following Solutions which allow NA cells to be differentiated from A cells:

1) A mixture of cosine-aniline blue» buffered at ph 4 with acetato buffer, 5M (Wood, 1963).

2) Giemsa solution» modified according to Pearse (1960).

3) Mailory trichromic stain.

The NA/A celi ratio was calcolateci from celi counts, using every tenth transverse section from thè whole gland of each specimen.

18 Boll Soc. Natur. Napoli - Voi 102 (1994-1995)

Results

The adrenal gland of all thè species studied appears almost elongated, surrounded by a thin connective envelope. The steroidogenic tissue is arranged into sinuous anastomosing cords of two celi rows separated by small blood vessels. The steroidogenic cells are prismatic, and in M. olivieri thè peripheral interrenal cells appear spongy, full of lipid droplets; thè nucleus, prevalently basally displaced, in A. boskyanus and M. olivieri , outlines thè shape of thè steroidogenic cords (Fig. 1-2).

Fig. 1 - Adrenal gland of Acanthodactylus boskyanus. The nucleus of thè steroido¬ genic cells, basally displaced, outlines thè shape of thè steroidogenic cords. Note thè cap of chromaffin tissue at thè head pole of thè gland. X270.

Fig. 2 - Adrenal gland of Mesalina olivieri. The steroidogenic cells appear spongy; note thè thin chromaffin ribbon almost devoid of digitations. X270.

In L. longicaudata and M. olivieri thè chromaffin tissue is located at thè head pole of thè gland, forming a cap, and extends also on thè dorsal

Capaldo A. et al » The adrenal gland oflìzards

margin in a thin discontinuous ribbon, almost devoid of digitations in M. olivìeri (Fig. 2), present instead in L. longìcaudata (Fig. 3). In A. boskyanus and A. pardalis (Fig, 1-4), thè chioma! fin tissue moreover is arranged in

Fig. 3 - Adrenal gland of Latastia longìcaudata . The chromaffin tissue is arranged in a cap at thè head pole of thè gland, and in a thin dorsal ribbon sendìng digitations between thè steroidogenic corda. XI 10.

Fig. 4 - Adrenal gland of Acanthodactylus pardalis. The chromaffin tissue is located at thè head pole and extends in a dorsal ribbon with many digitations into thè steroidogenic parenchyma. X 1 10.

a cap at thè head pole and in a dorsal ribbon sending digitations between thè steroidogenic cords; thè ribbon gets to thè ventral pole in A. boskya¬ nus. In E. strauchi (Fig. 5) thè chromaffin tissue is concentrated on thè poles of thè gland, on thè dorsal margin in a discontinuous ribbon with many digitations, and on thè ventral margin in small groups made up by NA and A cells: in E. pleskei (Fig. 6) thè ribbon is uniform and continuous.

In all thè species studied thè NA cells are located in thè outer layers of thè ribbon and of thè masses on thè poles, while A cells occupy thè

20

Boll. Soc . Natur. Napoli - Voi. 102 ( 1994

inner layers and thè digitations (Fig. 7, where NA cells appear dark grey and A cells light grey).

Fig. 5 - Adrenal gland of Eremias strauchi. Note thè chromafhn tissue arranged

at thè pole of thè gland. X270.

Fig. 6 - Adrenal gland of Eremias pleskei. Note thè chromafhn tissue arranged at thè poles of thè gland, at thè ventral margin, in small groups, and at thè dorsal margin in a uniform ribbon. Xl 10.

Fig. 7 - Adrenal gland of Acanthodactylus pardalis. Note thè noradrenaline cells (dark grey) in thè outer layers of thè dorsal ribbon, while thè adrenaline cells (light grey) occupy thè inner layers and thè digitations. X270.

21

Capaldo A. et al. ~ The adrenal gland of lizards

All thè species show some islets of A cells scattered between thè interrenal cords; they are of medium size and numerous in all thè species except M. olivieri, where they are few.

The different distribution of thè two types of chromaffin cells of thè adrenal gland influences thè quantitative ratios between NA and A cells. The ratios observed are thè following:

Mesalina olivieri	3.1/1
Latastia longicaudata	1.1/1
Acanthodactylus boskyanus	0.6/1
Acanthodactylus pardalis	0.5/1
Eremias strauchi	0.4/1
Eremias pleskei	0.2/1

Discussion

The study of thè morphology of thè adrenal gland in thè species examined shows two main arrangement patterns of thè chromaffin tissue: thè first exhibited by M. olivieri and thè second generally observed in thè other species.

M. olivieri shows a higher degree of separation between thè chromaffin and thè steroidogenic tissue; thè former, in fact, is concentrated at thè head pole to form a cap and extends on thè dorsal margin of thè gland in a thin, discontinuous ribbon, devoid of digitations; there are, moreover, few chromaffin islets interspersed in thè steroidogenic parenchyma.

In thè other species, instead, there is a higher degree of admixing of thè two types of tissue; thè chromaffin tissue is located at thè head pole and extends in a dorsal ribbon sending digitations between thè steroidoge¬ nic cords, as in L.longic andata, in A.pardalis and in A.boskyanus (in thè latter thè ribbon gets to thè other pole of thè gland); is arranged in E. pleskei in a continuous dorsal ribbon with many digitations, while in E. strauchi forms a discontinuous ribbon extending from thè head to thè ventral pole, and shows some clusters of chromaffin cells on thè ventral margin of thè gland. In all these species, moreover, many islets are inter¬ spersed in thè interrenal tissue.

These two different arrangement patterns of thè chromaffin tissue influence thè quantitative ratios between NA and A cells; this ratio is high, in fact, in M. olivieri (3.1/1), while is around 1/1 in thè other species.

According to thè results of thè preceding studies on thè morphology of thè adrenal gland of Squamata, we can suggest that there is a connec¬ tion between thè phylogenetic position of thè species, thè relationship

22

Boll. Soc. Natur. Napoli - Voi. 102 (1994-1995)

between thè chromaffin and thè steroidogenic tissue, and thè quantitative ratio between NA and A cells: in thè older species, in fact, thè adrenal tissues are more separated, and thè NA/A celi ratios are high; in recently evolved species, instead, these ratios are low, around 1/1 or less, and thè two tissues show a higher degree of admixing (Laforgia & Varano, 1982; Laforgia et al. 1982, 1983, 1985, 1990, 1991, 1993 a,b; Varano & Laforgia, 1982, 1988, 1991; Capaldo et al. 1991).

Therefore, according to our observations, we can suggest that M. olivieri should be considered an ancient species, while thè other species should be considered of more recent origin.

Particularly interesting is thè similarity of thè ratios shown by A.pardalis , A.hoskyanus (0.5/1 and 0.6/1) and A.erythrurus , examined in a preceding study (Capaldo et al. 1991). This species, in fact, with a NA/A celi ratio of 1.3/1 shows a relatively recent origin, such as thè other two species.

Phylogenesis and taxonomy within thè family Lacertidae have been carefully examined by Arnold (1989, 1993) and Bòhme & Corti (1993). Arnold, in particular, in a study on thè phylogeny and biogeography of thè Lacertidae, divides this family into two sections: a holophyletic, mainly African group with advanced forms extending into Eurasia, and a primi¬ tive Palearctic and Orientai assemblage that is very probably paraphyletic. Advanced taxa occurr in mainly xeric habitats; a group of increasingly advanced forms occupy thè Ethiopian region, as thè genus Latastia, while thè most advanced occurr in thè Saharan area and in thè drier parts of Eurasia, as thè genus Eremias, Acanthodactylus and Mesalina.

In particular, Arnold (1989) on thè basis of niorpho logie al features, associates thè genus Eremias (occurring prevalently in thè Ethiopian re¬ gion) with more advanced forms as Acanthodactylus and Mesalina, rather than consider it a “sister group" of thè genus Pedioplanis.

The homogeneity of thè Ethiopian and Saharo-Eurasian clade (except M.cuneirostris) was confirmed with karyological studies on thè localiza- tion of thè nucleolus organizer (NOR) that, in all thè species belonging to this group is of thè MS type (medium-small NOR-bearing machrochromo- some) (Olmo et al. 1991, 1993).

Our observation on thè adrenal gland of A.pardalis, A.hoskyanus , E.strauchi, E.pleskei and L.longic andata demonstrates that thè relation- ships between thè steroidogenic and thè chromaffin tissues, and thè quan¬ titative NA/A celi ratios are relatively homogeneous; M. olivieri, on thè contrary, is quite different from these species for both thè high NA/A celi ratio (3/1) and thè distribution of thè chromaffin tissue.

Capaldo A. et al - The adrenal gland of lizards

23

REFERENCES

Arnold, E.N. 1989. Towards a phylogeny and biogeography of thè Lacertidae: relationships within an Old-World family of lizards derived from morphology. Bull Br. Mus. nat. hist. (ZooL), 55 (2): 209-257.

Arnold, E.N. 1993. Phylogeny and thè Lacertidae. In: Lacertids of thè Mediterra- nean region (Ed. by E.D. Valakos, W.Bòhme, V. Pérez-Mellado), pp. 1-16. Athens, Bonn, Alicante: Hellenical Zoological Society.

Bòhme, W. & Corti, C. 1993. Zoogeography of thè lacertid lizards of thè western Mediterranean basin. In: Lacertids of thè Mediterranean region (Ed. by E.D. Valakos, W.Bohme, V. Pérez-Mellado), pp. 17-33. Athens, Bonn, Alicante: Hellenical Zoological Society.

Capaldo, A., Laforgia, V., Varano, L., Putti, R., Cavagnuolo, A. 1991. Distributive patterns of chromaffin cells in thè adrenal gland of reptiles belonging to family Lacertidae. In: Symposium on thè evolution of terrestrial vertehrates (Ed. by G. Ghiara), pp. 431-437. Modena: Mucchi.

Chester Jones, I. 1957a. Comparative aspects of adrenocortical-neurohypophyseal relationships. In: The Neurohypophysis (Ed. by H. Heller). New York: Acade- mic Press.

Chester Jones, I. 1957b. The adrenal cortex . Cambridge: Cambridge University Press.

Coupland, R.E. & Hopwood, D. 1966. The mechanism of thè differential staining reactions for adrenaline and noradrenaline Storage granules in tissues fixed in glutaraldehyde. J. Anat., 100: 227-243.

Gabe, M. 1970. The adrenal. In: Biology of Reptilia (Ed. by C.Gans), pp. 263-313. New York: Academic Press, London.

Gabe, M. & Martoja, M. 1961. Contribution à l’histologie de la glande surrénale des Squamata. Archs. Anat. Microsc., 50: 1-34.

Ghosh, A. 1977. Cytophysiology of thè avian adrenal medulla. Int.Rev.Cytol. , 49: 253-284.

Laforgia, V. & Varano, L. 1978. The influence of thè interrenal steroidogenic tissue on thè chromaffin cells of thè adrenal gland of Lacerta s. sicula Raf: effects of thè ACTH administration during thè winter. Celi and Mol. Biology., 23: 379-390.

Laforgia, V. & Varano, L. 1982. Morphology and distribution of thè chromaffin cells in thè adrenal gland of thè Cordylidae. A comparative study. J. Morph., 171: 79-88.

Laforgia, V., Varano, L., Cavagnuolo, A., Putti, R. 1982. Ulteriori osservazioni sulla morfologia della ghiandola surrenale degli Squamati. I Geconidi e gli Iguanidi. Boll. Zool., 49 (suppl): 101.

Laforgia, V., Varano, L., Cavagnuolo, A., Putti, R. 1983. La surrenale degli Aga- midi: studio morfologico in alcune specie. Atti XXXIX Conv.Naz.Soc.lt. Anato¬ mia, 430-431. Trieste.

Laforgia, V., Varano, L., Putti, R., Cavagnuolo, A., Capaldo, A. 1985. Morfologia della ghiandola surrenale in alcune specie della famiglia Lacertidae: uno studio comparativo. Atti XX Cong.Soc.It.Istoch . , 164. Milano.

Boll Soc . Natur. Napoli - Voi 102 ( 1994-1995 }

24

Laforgia, V., Varano, L., Capaldo, A., Cavagnuolo, A., Putti, R. 1990. Comparative morphology of thè adrenal gland in some species belonging to thè family Lacertidae. Amphibia-Reptilia , 11: 123-130.

Laforgia, V., Varano, L., Capaldo, A., Putti, R., Cavagnuolo, A. 1991. Comparative morphology of thè adrenal gland in selected species of thè genus Podarcis. Amphibia-Reptilia, 12: 153-160.

Laforgia, V., Capaldo, A., Varano, L., Putti, R., Cavagnuolo, A. 1993a. Compara¬ tive morphology of thè adrenal gland in some Mediterranean species of thè family Lacertidae. In: Lacertids of thè Mediterranean region (Ed. by E.D. Valakos, W. Bòhme, V. Perez-Mellado, P. Maragou), pp. 35-42. Athens, Bonn, Alicante: Hellenical Zoological Society.

Laforgia, V., Capaldo, A., Sciarrillo, R., Cavagnuolo, A., Putti, R. 1993b. Fine structure of thè adrenal gland in some Scincidae. Atti 7th O.G.M. S.E.H., 87. Barcelona, Spain.

Olmo, E., Odierna, G., Capriglione, T., Caputo, V. 1991. A karyological approach to thè systematics of Lacertidae (Reptilia, Sauria). Rev. Esp. Herp., 6: 81-90.

Olmo, E., Odierna, G., Capriglione, T. 1993. The kariology of Mediterranean lacertid lizards. In: Lacertids of Mediterranean region (Ed. by E.D. Valakos, W. Bòhme, V. Perez-Mellado, P. Maragou), pp. 61-84. Athens, Bonn, Alicante: Hellenical Zoological Society.

Pearse, A.G.E. 1960. Histochemistry theoretical and applied. Boston: Little, Brown and Co.

Saint Girons, H. 1976. Comparative histology of thè endocrine glands, nasal cavities and digestive tracts in anguinomorph lizards. In: Morphology and Biology of Reptiles (Ed. by A d'A. Bellairs and C.B. Cox.), pp. 205-216. London: Academic Press.

Tramezzami, J.H., Chiocchio, S. and Wasserman, G.F. 1964. A technique for light and electron microscopie identification of adrenaline and noradrenaline sto- ring cells. Cytochemistry , 12: 890-899.

Varano, L., Della Corte, F., Galgano, M. 1969. Brevi note sull'ultrastruttura delle cellule a catecolamine di Lacerta s. sicula Raf. Atti Soc. Pel. Sci. Fi. Mat. Nat., 15: 39-44.

Varano, L. & Laforgia, V. 1976. Struttura e ultrastruttura dell'interrenale della lucertola (Lacerta s. sicula Raf.) in esemplari primaverili normali e dopo somministrazione di ormoni estrogeni e di reserpina. Arch. hai. Anat. Em- briol., 81: 245-262.

Varano, L. & Laforgia, V. 1982. La morfologia della ghiandola surrenale degli Squamati: uno studio comparativo. Atti XXXVIII Conv. Soc. It. Anatomia, 407-408. Roma.

Varano, L. & Laforgia, V. 1991. Evolutionary trends in thè adrenal gland of reptiles. In: Symposium on thè evolution of terrestrial vertebrates (Ed. by G.Ghiara), pp. 291-303. Modena: Mucchi.

Wassermann, G.F. & Tramezzami, J.H. 1961. Complete separation of adrenaline and noradrenaline secreting cells in a snake's adrenal gland. Acta Physiol. Latino-Am., 11: 148-149.

Capaldo A. et al. - The adrenal gland of lizards

25

Wasserman, G.F. & Tramezzami, J.H. 1963. Separate distribution of adrenaline and noradrenaline secreting cells in thè adrenal of snakes. Gen. Comp. Endocri¬ no l, 3: 480-489.

Wood, J.G. 1963. Identification of and observations on epinephrine and norepine- phrine containing cells in thè adrenal medulla. Am. J. Anat., 112: 285-304. Wright, A. & Chester Jones, I. 1955. Chromaffin tissue in thè lizard adrenal gland. Nature , 175: 1001-1002.

Wright, A. & Chester Jones, I. 1957. The adrenal gland in lizards and snakes. J.

Endocr., 15: 83-99.

Boll Soc . Natur . JVapoft - Vo/. Ì02 (1994-1995): 27-34

21

Mean sea level and seismic warning

Palumbo F. and Rapolla E.

Dipartimento di Geofìsica e Vulcanologia, Università degli Studi di Napoli Federico II, Napoli, Italy

Key words: mean sea level, hydrographic levelling, seismic precursor.

Abstract. By means of hydrographic levelling we have proved that thè ground of Trieste is unstable, evidencing a significant oscillation of thè ground level in correspondence to thè 1976 intense earthquake of Friuli. The geodetic measurements are shown to he reliable earthquake ’s precursors and to provide a long term alarm System. The results have indicated that thè tide gauge records of Trieste cannot be assumed as levelling reference point for MSL studies as widely thought since yet.

Riassunto. Il livello medio del mare e la sorveglianza sismica.

Mediante livellazione idrografica è stato dimostrato che il suolo di Trieste è instabile, evidenziandone una sensibile oscillazione in corrispondenza dell'in¬ tenso terremoto che colpì il Friuli nel 1976. Viene altresì provato che le misure geodetiche possono essere affidabili precursori dei terremoti e utile strumento nella sorveglianza sismica. I risultati hanno mostrato che le registrazioni mareo- grafiche di Trieste, in quanto rilevata in una zona instabile, non possono essere assunte quale riferimento per gli studi sul livello medio del mare come si pensava fino ad oggi.

Introduction

Ground level deformations indicate reliable precursors of seismic and volcanic events (Zadro, 1992).

Ground level variations are determined both by means of repeated geometrie measurements and hydrographic levelling. Since mean sea level (MSL) data are available over thè Adriatic sea and allow thè hydro¬ graphic levelling, we have checked, by means of thè tide gauge records, whether thè large earthquakes of May-September 1976 in Friuli were preceeded by an oscillation of thè ground. The results have confirmed what excepted and indicated that thè ground in Trieste is unstable.

Received 25.4.94, accepted 27.2.96

28

Boll . Soc . Matur. Napoli - Voi. 102 (1994-1995)

Data and Methods

Changes in thè height of thè leveling mark are usually checked by IGM (Istituto Geografico Militare). Althought thè determinations are very accu¬ rate, they are expensive, require long operation times, and are observed only over long periods. Furthermore, thè observations are more difficult to make on islands at a great distance from thè land.

Some italian coast lines and islands are volcanic and show bradyseism movements. Reference to sea level might thus provide a simpler method of following thè movements of thè earth’s crust. Moreover, modern tech- nology provides equipment for reai time monitoring, transmission, recep¬ tion, and computation of data in a centrai observatory.

At Phlegrean Fields volcanic area (Naples) following a renewal of rapid uplift toward thè end of 1969, changes in thè height of thè region above MSL were closely monitored using thè usuai methods of trigonome- trical surveying between reference points in thè rising region and others in more stable areas and also by thè established but less common method of tide gauges (Vantroys, 1958). Clearly, thè latter method provides thè height of a fixed point on thè tide gauge above sea level directly, but considerable reduction is required to relate this instantaneous (or even daily) mean sea level to thè internationally adopted level. To compensate for this additional effort, thè fidai records provide a continuos measure of thè height, whereas trigonometrical methods give values only at thè in- stant of observation. Moreover, trigonometrical methods require nume- rous intermediate stations and are therefore expensive and laborious. Corrado and Palumbo (1974) computed bradyseism movements at Poz¬ zuoli with an accuracy comparable to that obtained by means of trigono¬ metrical levelling carried out every month from 1970 to 1973. The success was attributed to thè fact that thè stable area (Naples) and thè bradysei- smic area (Pozzuoli) were only 10 km apart and thè tide gauge stations were all located in thè Bay of Naples. The removai of thè similar marine and meteorological contributions at thè two stations was obtained by simple differentiation of thè simultaneous sea level at thè two stations.

We now extend thè investigation to thè station of Trieste. This because of thè importance of thè tide gauge records of such a station for all MSL studies in thè Mediterranean. We thought that thè stability of Trieste tide gauge bench mark was reasonably to be checked considering a time interval including 1976 when thè area was interested by two intense earthquakes (M>4). In order to search for stable sites nearby Trieste we have examined thè available MSL records of tide gauge of northern Adria-

29

Palumbo F. and Rapolla E. - Seismic warning

tic stations reported in Fig. 1 (Koper, Dubrovnik, Split Rt Marjana, Split Harbour, Rovinj, Bar) for a time interval including 1976 (PSMSL). We

REGIONE BALCANICA

f'*' Scila 1 ; 10000000

Figure 1 - Northern Adria- tic stations: TRIESTE (45°39' N, 13°45’ E), KO¬ PER (45°33’ N, 13°44’ E), ROVINJ (45°05’ N, 13°38' E), DUBROVNIK (42°40’ N, 18°04' E), SPLIT RT MARJANA (43°30' N, 16°23' E), BAR (42°05' N, 19°05' E), SPLIT HARBOUR (43°30' N, 16°26’ E).

30

Boll Soc. Natur. Napoli - Voi 102 ( 1994-1995 )

have thus examined thè differences between thè yearly values of MSL observed in Trieste provided by thè Istituto Talassografico CNR of Trieste and thè corresponding annual values of thè MSL averaged over all thè other stations found to be stable.

Results and discussion

The differences between thè mean annual values of MSL in some selected stable stations (Koper, Dubrovnik, Split Rt Marjana, Split Har- bour, Rovinj and Bar) are reported in Fig. 2. This figure indicates that all

YEARS ' YEARS

Figure 2 - Differences between yearly data of MSL recorded in each pair of fide gauge stations excluding Trieste. (DV=Dubrovnik, ROV=Rovinj, SPH=Split Har- bour, SPR=Split Rt Marjana, KO= Koper, Bar- Bar , TS=Trieste). SOLID LINE= 5yr running means DOTTED LINE= observed values.

Palumho F. and Rapolla E. - Seismic warning

Figure 2 eontinued

thè selected stations show not signifìcant trends and long period oscilla- tions whereas they exhibit small and randomly distributed interannual fluctuations related to thè small different locai response of MSL to meteo- rological forcing, indicating thus thè ground level stability of thè selected stations. Many stations such as Venice and Bakar (Fig. 3) show large time variations mostly related to thè locai ground unstability and were thus not considered in thè present investigation.

Tab. 1 reports thè yearly data of MSL in all thè examined stations and thè differences between thè MSL observed in Trieste and thè average of thè MSL of all thè other stable stations (Fig. 4).

The data show a signifìcant symmetric oscillation of thè ground in Trieste from 1971 to 1982 with an amplitude around 2 cm. The range of

32

Boll Soc . Natur, Napoli - Voi 102 ( 1994-1995 )

Figure 3 - Bakar’s station shows large time variations related to thè locai instabi-

lity of thè ground.

this long term variation, being much higher than thè mean interannual variability (0.4 cm), indicates thè heigh statistical significance of thè for- mer. Such a result may be related to thè contemporary vertical oscillation of thè ground with an amplitude of 7 cm in thè Friuli epicentral area, ascertained by thè National Levelling Service (IGM).

In conclusion, thè present paper has shown thè utility of MSL studies which yielding continuos records of MSL differences between thè tide gauge bench marks in tectonic stable and unstable areas allow to obtain information on thè vertical movements of thè ground in tectonic unstable areas and hence provide a simple method to detect precursors of large earthquakes. Moreover thè results emphasize thè utility of warning thè ground level deformation throught thè hydrographic levelling in our

Pai-ambo F< anJ Rapolla E - Séismic warnìng

Tàble 1 - Yearly data of MSL (in cm) in thè examined stations and thè differences between thè MSL in Trieste and thè average of thè MSL of all thè other stable stations.

	DV	ROV	SPH	SPR	KO	BAR	TS	TS-MED
1967	15.1	14.0	14.6	9.5	10.5	9.8	9.4	-2.9
1968	16.8	15.5	15.8	ILO	11.5	12.4	11.4	-2.4
1969	21.6	19.6	21.1	16.4	9.5	17.4	15.3	-2.3
1970	20.0	18.1	19.1	14.4	13.0	16.1	13.8	-3.0
1971	18.5	16.2	17.3	12.4	11.9	14.3	11.6	-3.5
1972	16.0	15.6	15.3	11.0	11.1	11.9	10.5	-3.0
1973	13.8	10.2	11.9	7.4	8.5	10.3	7.0	-3.4
1974	16.0	14.2	15.1	10.4	12.5	13.0	10.5	-3.0
1975	13.2	11.8	12.3	6.8	8.5	9.8	9.0	-1.4
1976	13.7	12.2	12.8	9.1	9.1	11.2	9.7	-1.8
1977	14.6	15.5	14.5	9.2	12.0	10.3	12.1	-0.6
1978	17.5	17.2	17.2	12.2	13.1	13.0	14.1	-0.9
1979	20.6	19.3	19.8	14.1	12.7	17.0	16.0	-1.3
1980	19.3	16.8	18.7	13.2	13.2	16.4	13.3	-3.0
1981	19.2	17.1	17.2	13.3	13.4	17.4	13.9	-2.4
1982	18.0	15.3	17.0	12.1	12.6	17.4	11.7	-3.5
1983	16.5	14.7	14.5	9.2	11.8	13.9	11.2	-2.2
1984	19.6	18.0	18.5	12.5	14.5	17.3	14.4	-2.3

■ YEAR

Figure 4 - Differences between thè MSL observed in Trieste and thè average of thè MSL of thè other stable stations. SOLID LINE- 5yr mnning means DOTTED

I. INI: observed values.

34

Boll Soc . Natur. Napoli - Voi 102 ( 1994-1995 )

country being thè majority of seismo-volcanic Systems locateci along thè Coastal lines.

Finally thè present results warn us to consider thè tide gauge records of Trieste as reference data for MSL studies as widely assumed since yet.

REFERENCES

Corrado, G., A. Palumbo, 1974. Recente evoluzione del bradisisma flegreo. Boll Soc. Nat. Napoli, 83: 1-12.

Mazzarella, A., A. Palumbo, 1982. Mean sea level variations and their practical applications. Jour. Geoph. Res., 87: 4249-4256.

Mazzarella, A., A. Palumbo, 1989. Recent changes of mean sea level in thè Medi- terranean area. Boll. Oc. Teor. ed Appi. 7: n. 4, 285-293.

PSMSL, 1995. Montly and annual mean heights of sea level. Permanent Service for mean sea level, Institute of Oceanographic Science, Birkenhead U.K.

Vantroys, L., 1958. Note sur l'utilisation de la surface libre des mers dans les operations de nivellement. Bull. Geod. Fr., 52: 80.

Zadro, M., 1992. Tilt and strain variations in Friuli, NE-Italy after thè 1976 ear- thquake. “Earthquake Prediction”, M. Dragoni and E. Boschi Edito rs.

Boll. Soc. Natur. Napoli - Voi. 102 (1994-1995): 35-41

Mean sea level fluctuations and volcanic activity

Palumbo F. and E. Rapolla

Dipartimento di Geofisica e Vulcanologia,

Università degli Studi di Napoli Federico II, Napoli, Italy

Key words: mean sea level, earthquake triggering, volcanic warning.

Abstract. Significant lunar M2 tidal term, related to thè corresponding lunar oscillation in mean sea level (MSL), has been identified in historical series of earthquakes recorded around Vesuvius, Phlegrean Fields and Hawaii volcanoes. Both thè amplitudes and thè signifìcance level of such tidal term, increasing with thè volcanic activity, are proposed as precursors of eruptions.

Riassunto. Oscillazioni del livello medio del mare e attività vulcanica.

Nelle serie storiche dei terremoti registrati intorno al Vesuvio, ai Campi Flegrei ed ai vulcani delle Hawaii è stato identificato un significativo termine mareale lunare M2. Tale termine risulta collegato alla corrispondente oscillazione del livello medio del mare (MSL).

Inoltre sia le ampiezze che il livello di significatività di tale termine aumen¬ tano con il livello di attività vulcanica e perciò vengono proposte come precursori delle eruzioni vulcaniche.

Introduction

The possible influence of mean sea level (MSL) variations on volcanic activity has been hypothesized (McGuire, 1992), but quantitative data and thè statistic level of signifìcance of thè correlation is rarely reported in thè literature. The present paper searches for thè relationship between thè MSL variations and thè seismo-volcanic activity and discusses also thè reliabilìty of thè obtained results.

In more detail, thè paper tries to answer to thè following questions: May MSL variations:

a) help in warning volcanic and seismic events? b) trigger thè volcanic activity?

Received 25.4.94, accepted 26.2.96

36

Boll, Soc, N&iur, Napoli - Voi 102 (1994-1995)

Data

The data analysed are:

Phlegrean Fields (40° 49' N latitude; 14° 04' E longitude). The available shocks were recorded by thè locai seismic network. Two separate groups of data were analysed, according to two different level of volcanic activity: n° 1064 shocks with magnitude less than 2 were recorded from 1975 to 1980 (Palumbo, 1985) during a very low volcanic activity period; n° 100 shocks with magnitude higher than 2 were recorded during thè high volcanic activity (bradyseismic) period, from 1983 to 1984 (Luongo et ah, 1996).

Vesuvius (40° 49' N latitude; 14° 24' E longitude). Imbò (Imbò, 1959) has computed hourly indices of seismic activity at thè Vesuvius during thè interval between March 13 and Aprii 3, 1944, corresponding to thè last empii on of thè volcano. Two different groups of these high quality data corresponding, respectively, to a lower and a higher level of volcanic activity were separately analysed. Another set of 694 shocks from 1972 io! 993 (Luongo et al, 1996) having magnitude higher than 2, was also analysed. The latter were observed by thè locai volcanic seismic network during thè present quiescent state of thè volcano.

Hawaii volcanoes : to compare thè Phlegrean Fields and Vesuvius results with those relative to other different volcanic areas n. 31130 locai shocks recorded by thè Hawaii Volcano Observatory (NOAA) during 1970 1976; (17°-22° N, 152°-157° W), were also analysed.

Methods of analysis

For thè determination of thè M2 tidal term in thè seismic data an adjusted version of thè Chapman Mi] ler Method (CMM) was applied. The M2 term in MSL was obtained by means of routine tidal analysis.

When thè geophysical data are poor and randomly distributed, thè CMM requires some simple adjustments. The index V (lunar phase inte- ger) of thè day of each event can be computed according to Sugiura and Fanselau (1966) or to other techniques (Malin and Chapman, 1970). The events can be entered in a matrix (24 X 12): thè 24 columns correspond to thè 24 mean locai solar hourly intervals; thè 12 rows correspond to thè

Palumho F. and Rapolla E. - Tidal triggering 37

lunar index v\ One can look then for correlation with thè peak of thè actual tide and with spectral components.

In thè first case, since t = t-v' (t and t denote respectively thè solar and lunar time measured from locai lower transit of thè mean sun and moon), one can obtain thè distribution of thè events according to lunar hours and look for correlation with thè time series of thè actual tidal stresses.

If one searches for spectral determinations at least one information for each row of thè matrix is required. The 12 suro - sequences can thus be worked according to Malin and Chapman (1970) that prò vide also thè vector probable error (v.p.e.) of each determination.

If N is thè number of days of thè entire series of data examined, thè number of days for each sum sequence is N/12. The amplitudes of thè diurnal waves A in each sum-sequence are N/12 times thè actual ones Aa, i.e. A=Aa/12. If Ns is thè N total number of signals occurred during thè time interval of thè series of data, thè mean value of thè figures entered in thè matrix is Ns (12X24). The ratio

AaN/12 _ Aa NS/(12X24) NS/(NX24)

indicates thè value of thè amplitude given by thè routine program. One can obtain then thè percentage of thè diurnal amplitude of thè single wave with respect to thè hourly mean value of thè geophysical signal.

The routine program written by Malin and Chapman (1970) provides also thè vector probable error of each determination, that is thè scattering of thè individuai harmonics calculated from separate nonoverlapping su- bgroups of data from thè one obtained from thè whole series of data. A determination is significant at 0.05 level when thè amplitude is 2.08 times its probable error.

Results

In thè time domain it can be noted that thè two recent events of ground uplift at thè Phlegrean Fields during 1970 and 1982 were found to be preceeded by an anomalous large increase of Oceanie loading (Pa¬ lombo, 1985). A significant covariance of thè annual cycle of Oceanie loading with thè number of shocks was also ascertained.

38		Boll Soc. Natur. Napoli - Voi. 102 (1994-1995)
In thè frequency domain, thè determined main tidal M seismic data are schematically reported in Table 1. Table 1	2 terms in
VOLGANO Name	AMPLITUDE in thè seismic M2term	TIME OF MAXIMUM OCCURRENCE (hr)	LEVEL OF VOLCANIC ACTIVITY	VENT
Phlegrean	10 ± 4.3	3.5	low	closed
Fields	72 ± 16	3.0	high
1982-83
Vesuvius	51 ± 23	4.3	high
1944	85 ± 32	9.5	very high	open
Vesuvius	21+9	0.1	quiescent	closed
1972-93
Hawaii	3 ± 1	3.7	all	open

The amplitudes indicate thè percentage of thè lunar semi-diurnal amplitudes with respect to thè mean hourly number of shocks. For thè data of Vesuvius 1944, thè amplitudes are thè percentage with respect to thè hourly mean value of thè seismic index computed by Imbò (1959). The times of maximum occurrence are thè lunar hours after thè lower locai lunar meridian transit.

The MSL amplitude (1 1.6 cm) and thè phase angle (260°) of thè lunar M2 component have been determined, from thè tide gauge records of thè stations located in thè Bay of Naples, according to routine tidal analysis.

It is worth noting that all determinations were obtained at a signifi- cance level higher than 0.05. Similar results have been obtained for other Principal tidal terms.

Discussion

Question a) May MSL variations help in warning volcanic and seismic events?

Vertical movements of thè ground are reliable precursors of seismic and volcanic activity. Since thè great majority of active volcanoes and

Palumbo F, and Rapolla E. - Fidai triggering

39

seismic areas are located along Coastal lines or on islands thè hydro- graphic levelling is more suitable to detect thè vertical movements of thè ground that are volcanic and seismic precursore.

If one or more mareographic stations are located in thè stable area nearby another tide gauge station located in an unstable seismic or volca¬ nic site, thè hydrographic levelling may be obtained by computing thè differences between thè MSL at thè two stations. Such differences can give a significant estimation of thè vertical movements of thè mareographic bench mark located in thè unstable station.

Successful results were obtained in thè Bay of Naples by thè use of hydrographic levelling between thè unstable site of Pozzuoli and thè stable one in Naples (Corrado and Palumbo, 1974) providing nowaday routine measure of thè vertical movements of thè ground in that unstable area.

To correlate MSL variations to thè ground level ones, statistica! ap- proach requires thè availability of numerous fluctuations. The tidal varia¬ tions seem thus to be thè most appropriate data sets. Tidal terms may be obtained by routine tidal analysis or by means of thè more sophisticated CMM summarized above.

Coherent non tidal long term oscillations in MSL and ground level may also be easly determined owing to their large amplitudes. The con- temporary tidal (and non tidal) components that would be present in both thè MSL and thè ground level data would indicate thè possible physical relationship between themselves on one hand and let identify in thè ground level variations induced by MSL ones, a signal of seismic hazard on thè other.

Question b) May MSL variations trigger thè volcanic activity?

The Neapolitan area, where about 3 million of people live, is located between two active volcanoes (Vesuvius and Phlegrean Fields). In Europe this is thè area with thè highest volcanic risk; thè magmatic chambers of both volcanoes lay below thè Bay of Naples. If a chaotic structure of these volcanic systems is identified, a possible response of thè volcanic activity to an external excitation such as thè MSL variations may be expected.

A fractal approach to both volcanic systems was successfully applied and showed that thè volcanic seismicity can be much better described as a fractal structure than as an euclidian one (Luongo et al., 1996 a,b).

Both volcanic systems are at present in a self-organized criticai state in which a minor event like shess from Oceanie tidal effeets (Chiaruttini, 1976) and from thè body Tide, may start a chain reactions that can affé et any number of elements in thè System.

The small amplitude at Hawaii may be related both to thè low visco- sity of thè lava and to thè open vent at this volcano. The results for thè Phlegrean Fields and for Vesuvius show that thè lunar amplitudes and their significance increase with thè level of volcanic activity.

The significant lunar tidal terni M2, obtained in thè different examined volcanic areas, indicate that thè lunar triggering effect is related both to thè structure of each volcano and to thè level of their volcanic activity. This would show that thè tidal M2 term may be assumed as a seismic precursor and hence as an eruption hazard indicator. From a physical point of view when thè magma is inside thè magmatic chamber, thè tidal induced oscillation do not interact with thè solid structure of thè volcanic building and does not thus stress them. When thè level of activity is high there is reasonably a magmatic intrusion in thè rocks. In this case thè tidal oscillation of thè fluid magma will interact with thè surrounding rocks stressing and microfracting them and hence causing thè tidal related earhquakes evidenced by thè present analysis at Vesuvius and Phlegrean Fields during high level of volcanic activity.

Acknowledgements

The present work has been developped in thè framework of thè EC research project EV5V-CT93-0266 “ Relative sea level changes and extreme floading events around European coasts ”.

REFERENCES

Chiaruttini, C., 1976. Tidal load on thè Italian peninsula. Geophys. J.R. astr. Soc.) 46: 773-793.

Corrado, G., A. Palumbo, 1974. Recente evoluzione del bradisisma flegreo. Boll. Soc. Natur., Napoli, 83: 1-9.

Imbò, G., 1959. Oscillazioni a periodo semidiurno lunare dell'attività eruttiva vesuviana. Atti Vili Conv. Annuale Ass. Geof. Italiana.

Luongo, G., A. Mazzarella and A. Palumbo, 1994. A fractal approach to clustering of 1983-84 seismicity in thè Campi Flegrei Caldera, Southern Italy. J. Geoph. Res. Leti, in thè press.

Luongo, G., A. Mazzarella and A. Palumbo, 1996. On thè self organized criticai state of Vesuvio volcano. Jour. Volc. and Geoth. Res. 70: 67-73.

Malin, S.R.C., S. Chapman, 1970. Geophys. J. R. Astron. Soc., 19, 15.

Palumbo F. and Rapolla E. - Tiàal trigger ing

41

McGuire, W.J., 1992. Changing sea levels and erupting volcanoes: cause and effect. Geology Today, 8: (4), 141-144.

NOAA (National Oceanie Atmospheric Administration): U.S. National Geophysical Data Center Boulder Colorado Issue n. 80303.

Palumbo, A., 1985. Influence of external tidal and meteorological forces on thè bradyseismic phenomenon in Phlegrean Fields. Il Nuovo Cimento C, 8, 538- 551.

Palumbo, A., 1986. Lunar and solar tidal components in thè occurrence of ear- thquakes in Italy. Geophys. J. R. Soc. 84, 93-99.

Sugiura M., G. Fasenlau, 1966. NASA Report X 612-66-401, Goddard Space Flight Center, Greenbelt, Maryland.

Boll Soc. Natur. Napoli - Voi 102 (1994-1995): 43-47

43

Triassic estherids from Salerno (Southern Italy): preliminary notes

Barbera C., Lamagna R. e Signore M.

Dipartimento di Paleontologia, Università degli Studi Federico II, Napoli, Italy

Key words: Crustacea, Lagerstàtten, Triassic., South Italy

Riassunto. Esterie triassiche del salernitano (Sud Italia): Nota prelimi¬ nare.

Viene segnalata, per la prima volta, nel Norico del salernitano la presenza di Estendi ed in particolare di Isaura minuta (Von Zieten). I ritrovamenti sono avvenuti nella zona di Gittoni Vallepiana ed in particolare alle pendici di M. Pentirne nelle dolomie bituminose che per lungo tempo hanno fornito materiale sfruttato nelle minire di ittiolo. Queste dolomie si erano depositate in bacini anossici in ambiente marino come testimoniato anche dagli altri fossili rinvenuti (molluschi, vertebrati). La presenza di esterie in questi sedimenti, dato che queste forme sono sicuramente di acqua dolce fa ipotizzare l'esistenza nelle immediate vicinanze di terre emerse e di bacini fluviali ove questi organismi potevano vivere. In un appendice alla fine del lavoro è stata effettuata una messa a punto sistema¬ tica del taxon.

Abstract. In this paper there are thè preliminary considerations about thè findings of estherid crustaceans in thè Norian of Salerno area. The anoxic basin in which thè estherids have been found was marine environment, as confirmed by thè fossils of many fishes, but thè estherids were not marine beings. Thus, it may be that there were sweet water apports in thè shallow Coastal basin of thè area.

Introduction

The excavations in thè area around Salerno are known from thè beginning of this century. In thè first years of thè 1900s, in need for fossil fuels, mining facilities were built in many areas in Italy, searching for black shale limestone, in which fossil fuel may be found. In Northern Italy,

Received 23.6.95, accepted 26.1.96

44

Boll. Soc . Natur. Napoli - Voi 102 (1994-1995)

this activity was well developed, and thè “ittiolo” (= fish oil), thè fuel extracted from thè Triassic black rocks, was used in many ways, mainly in producing medicai remedies. To exploit this new field of production, in thè area around Salerno flourished many extraction facilities. The product was called “saurolo” (= reptile oil). Nowadays thè mines are abandoned, standing with their stony carrions, battered and crumbling from thè stri- kes of time. But in these areas, since thè beginning of thè century, were noted fossils. The main fossil being from these area are fishes. Primitive bony fishes with a stili heavy armor, not thè slender and swift forms living today. But along with fishes, were found many other animals. Among them, some small crustacean, thè estherids.

In this paper we outline thè ecology of this group, and thè possible environment in which they died.

Systematic and paleoecology

The small freshwater crustacean known as estherids are an order of thè subclass Diplostraca (see Appendix one), called Conchostraca. The specimens from Salerno area seem (at a first analysis) to belong to thè suborder Spinicaudata, but further studies (now in progress) are needed to determine thè superfamily.

9

Figure 1 - Specimens of I saura minuta, thè possible genus of thè Salern area. Note

THE SEXUAL DIMORPHISM X 5.

The sampled specimens have an anathomy similar to thè generai appearence of thè order Conchostraca. They show a shell composed of two valves, laterally compressed, that encloses thè soft body. The carapa- ces, formed by two distinct parts (Tasch, 1969), one chitinous and one

Barbera C. et al - Triassic estherids from Salerno

45

membranous, has well marked growth lines, that are different for thè various genera. The lines reflect successive moltings, i.e. in addition to thè ecdysis, thè animai adds a new line of growtth periferically to each valve. The number and thè width of these lines vary, according to thè genus, environmental condition, and sex. In fact, thè conchostracans show a well marked sexual dimorphism, that is very evident even in thè fossil speci- mens.

Although thè most of living conchostracans display various ornamen- tations, in thè specimens from Salerno we have not yet observed precise patterns of ornamentation. But this may be due to a bad fossilization of thè shells, as well as due to a heavy transport of thè shells themselves, that may have obliterated thè most visible features.

The orientation of thè carapace is easy. The umbonal area is thè dorsal part, and thè head of thè animai was about under thè umbo.

Sometimes, thè fossil conchostracans show soft parts. This is of great help, especially with thè eggs, since it separate clearly thè female from thè male. Sadly, this doesn’t seem thè case of thè specimens from Salerno, even though in one specimen (until now) it seems to be a cast of soft parts.

The main habitat of these crustaceans happen to be thè small, tempo- rary, alkaline inland ponds (Tasch, 1969). These ponds are never very large, and usually host only one species of a particular genus. The same thing occurs with fishes, hence thè cohabitation appears possible only when thè ponds in which they live became connected with some greater water bodies (e.g. rivers, or lakes) (Frank, 1988).

Anyway, thè conchostracans have been found even in spring water and in Coastal salt flats.

The paleozoic Conchostracan come from a marine environment, and are associated with many forms, included trilobites and ammonoids, as well as in thè coprolites of coelacanths (Tasch, 1969). Possibly, thè most recent forms began to live in thè estuarine Systems (as noted by thè Irish genus Limnestheria , from thè Carboniferous, in Tasch, 1969). According to fossil evidences, it seems that from thè Carboniferous thè conchostra¬ cans migrated to thè freshwater environment. This event may not have been at one time, but more possibly it happened in several “pulses” at different times.

From thè Mesozoic we find estherids in fresh waters. This may be a problem when we find them in strata that are usually marked as marine. And this seems to be thè case of thè Salerno area.

46 Boll. Soc. Natur. Napoli - Voi 102 (1994-1995)

The anoxic basins around Salerno

The samples collected around Salerno, especially from Giffoni Valle- piana include many fishes, some reptilian remains (mainly teeth of a possible thalattosaurid), and thè conchostracans. Now, thè strata from which thè samples come are made by black shale. This marks an anoxic bottom in a shallow Coastal basin. The evidences found in thè area show that thè anoxy was not Constant, but was alterrnated with normal salinity. Hence we may think that thè basin of which Giffoni was a part was interested by periodical current changes, that determinated thè anoxic condition. During thè “normal” stages life could be as in any other Coastal basin, but during thè anoxic stages thè prohibitive environmental condi¬ tion that settled down would have made life impossible. At this stage thè only life in thè basin was a cyanobacterial mat, that covered anything (as in other Lagerstàtten, e. g. Solnhofen, Holzmaden, Bolca, Pietraroia), allowing thè typical fossilization of thè Plattenkalke basins.

Aside from thè mechanicals of thè Plattenkalke basins, that are not thè objective of this paper, thè point is: how did thè estherids come in thè basin, and when ?

It may be possible that around thè area of thè basin there was many small islands, but for thè spreading of thè estherids any land mass may be useful. As noted before, thè conchostracan can live in any temporary pond. Another important fact is their eggs. The conchostracan eggs can hatch even after bad environmental conditions. Studies show how they batch at approximatively thè same time even when they have been kept dry, frozen, or moist (Tasch 1969). Moreover, eggs are laid at every ecdy- sis, attached to thè old exopodite cuticle. Thus there are many generations of eggs in a short time. And thè eggs can be transported by thè water as well as by wind.

Hence, if we hypotize that around thè anoxic basin there was even a small land mass, surely it would have rain on it, and thè draining waters brought thè conchostracans from their ponds to thè sea, where they died, or arrived already dead. This seems thè most acceptable explaination of freshwater forms in a marine basin, at least until more studies on these conchostracan may be done.

APPENDIX ONE: SySTEMATICS

In this Appendix (mainly from Tasch, 1969, and Reibl, 1962) is repor-

Barbera C. et al. - Triassic estherids from Salerno

47

ted thè systematics of Conchostraca found thè Salerno area, according to thè preliminary hypothesis.

Phylum Arthropoda Subphylum Crustacea Class Branchiopoda Subclass Diplostraca Order Conchostraca Superfamily Spinicaudata Family Isauridae Genus Isaura

Isaura minuta (Von Zieten, 1833)

The studied material appears to belong all to this species, but we are stili not sure about thè subspecies, since thè specificai marks have not yet been studied, and we are stili waiting for material.

Anyway in this this preliminar note is for thè first time marked a Continental species typical of German Upper Keuper in deposits of Sa¬ lerno area until now cnsidered of marine environment.

REFERENCES

Barthel K.W., Swinburne & Conway Morris S., 1990. Solnhofen. A study

in Mesozoic paleontology. 236 pp., Cambridge U.P.

Frank P.W., 1988. Conchostraca. Paleogeography, Paleoclimatology, Paleoeco- logy, 62: 399-403.

Reible P., 1962. Die Conchostrachen (Branchiopoda, Crustacea) der germani- schen Trias, N. Jb. Geol. Palaont., Abh: 114: 169; 244.

Tasch P., 1969. Branchipoda. In Moore R.C. (ed.) Treatise on Invertebrate Paleon¬ tology . (R) 4/1: 128-191.

Tasch P., 1987. Fossil Conchostraca of thè Southern Hemisphere and Continental Drift. Paleontology, Biostratigraphy and Dispersal. Geol. Soc. Am. Mem. 165: 1-290.

Boll Soc. Natur. Napoli - Voi 102 ( 1994-1995 ): 49-56

A miocenic reef near Catanzaro, Calabria, South Italy

Barbera C., Bilancio F., Giordano P.

Dipartimento di Paleontologia Università degli Studi Federico II, Napoli, Italy

Key words: Reef, Miocene, South Italy

Abstract. On this work we describe a miocenic reef with Porites and Tarbel- lastrea near Vibo Valentia (South Italy).

This reef of miocenic age lie on cristalline basement and is heteropic to sands with Clipeastrids. We describe a stratigraphic section made about 100 m fom this reef to clear thè relation between thè different formations of thè sequence.

Riassunto. Una scogliera miocenica presso Catanzaro (Calabria) Sud Italia.

Nel presente lavoro si descrive una scogliera miocenica affiorante presso Vibo Valentia. Da un rilievo geologico particolareggiato della zona si è ricostruita una successione stratigrafica dell’area ove sono state riconosciute le seguenti unità litologiche:

Graniti e grano dioriti, non estesamente diffusi nella zona in esame.

Scisti cristallini, formati da gneiss biotico-granatiferi.

Sabbie argillose verdi-grigiastre, poggianti su basamento cristallino, datate Miocene Medio.

Arenarie e sabbie grigie e giallastre, diffuse in tutta l’area, datate Miocene Superiore.

Argille grigie-azzurre, di spessore modesto, la loro età è Miocene Superiore (Messiniano).

Calcari biancastri datati Miocene Superiore (Messiniano): rappresentano, con la formazione precedente, la chiusura della trasgressione Miocenica.

Marne bianche (Trubi), si ritrovano esclusivamente lungo la fiumara Traeniti e segnano l'inizio della trasgressione Pliocenica.

Conglomerati sciolti, si rinvengono presso Vena Inferiore: sono del Quaterna¬ rio.

La scogliera si trova a circa 100 m. NE dalla sezione poggia sul basamento cristallino ed è costituita essenzialmente da Porites e Tarbellastraea.

Introduction

The upper Miocene of Calabria is represented by sandstones with Clypeastrids who pass to fossil coral reefs made by Porites and Tarbel-

Received 23.6.95 , accepted 26.1.96

50

Boll Soc . Natur. Napoli - Voi 102 (1994-1995)

lastrea. Those reefs are recognisable near Vibo Valentia (fig. 1). The reefs are built on metamorphic basement made by granite or gneiss or on thè grainstones with Clypeastrids and pass to grainstones with red Algae, Molluscs, Brachiopods, Echinid and larger Foraminifera.

Figure 1 - Location of thè area.

Among thè Authors that have studied this area, we note thè following: Seguenza G. (1879), Neviani A. (1887), De Stefani C. (1887), Cortese E. (1895), Checchia-Rispoli G. (1937), Zuffardi-Comerci (1937), Nicotera P. (1959), Chevalier (1967), Morelli et al. (1969), Ogniben L. (1973), Civetta I. et al. (1973), Fatton E. (1973), Ortolani F. (1975), Barbera C.-Tavernier A. (1988-1989), Saint Martin (1990), Demarq G. (1989-1990). A lot of them mark that corals building were present in this area.

Chevalier (1967) spoke about one miocenic reef near Vibo Valentia; he says that this reef is thè most important of thè area.

Barbera and Tavernier (1989) studied different sections near Vibo Valentia and described thè sequence who is made by pebbles, sandstones with Clypeastrids and covered by trasgessive sediments of thè late Pliocene.

51

Barbera C. et al. - A miocenic reef near Catanzaro, Calabria

■

Geological situation

The area is thè flattened top of a geological structure that represent thè slope toward thè sea in which we recognize an hydric radiai circula- tion with a torrential System. This kind of circulation make thè deep canyons that we can observe.

On thè paleozoic metamorphic basement we found a Cenozoic sedi- mentary sequence made by:

1) Sand with clays green-grey of various size in which we note a carbonious level with Ostreids and Ceritiids.

2) Sandstones and grey yellow sands of various size and with a differ- ent degree of cementation with a lot of different fossils (we recognise Clypeastrids, Pectinides, Heterostegineds, Brachiopods, Ostreids, Cheloni- ans, theeths of fishes and bones of sea mammals). On this sand we can found pieces of coralline biocostrutions similar to those found near C. Pioppo (near military airport of Vibo Valentia) The most important species present is Porites stratiformis .

3) Grey-blue clays without macrofossils that pass on thè upper part to “Tripoli“-like marls. Those clays contains many lignitiferous levels.

4) White-yellow dusty friable limestone, in horizontal banks which lie on thè grainstones with Clypeastrids and on thè clays that in tura are eteropic with thè grainstones.

All those levels are dated between thè Tortonian and Messinian.

The stratigraphic section

A stratigraphic section located near thè road between thè military airport of Vibo Valentia and thè village of Cessaniti in a location called “Stretti" has been studied in detail.

This section ranges from 450m to 508m above sea level. Their thick- ness is about 58m (Fig. 2).

We can observe essentially two kinds of lithotypes. At thè base of thè sequence we observe a sandy-clay formation with a lens of black clay very rich in Crassostrea gryphoides. Upon it we find sands and sandstones with a various degree of cementation, grey at thè base and becoming yellow toward thè top.

52

Boll Soc . Natur. Napoli - Fo/. 102 ( 1994-1995 )

Figure 2 - Field view of thè quarry.

1 - Granits; 2 - Grey-blue sands; 3 - level witrh very little Clypeastrids; 4 - Sands with Heteroste- ginids, Pectinids, Ostreids, internai cast of Gas- tropods and rare Clypeastrids; 5 - level with big Clypeastrids; 6 - Yellow sands without fossils.

Barbera C. et al - A miocenic reef near Catanzaro, Calabria

53

The reef

The reef that we can observe near Contrada Pioppo has a lenght of about Ikm and is fragmentarily exposed. Only in one place its lenght is about 200m and high about 35m. Outside thè reef we find bioclastic sands rich of Bryozoans, Echinids and Red Algae.

Fauna composition

The exposed pari of thè reef is made by:

Porites stratìformìs (Seguenza)

Porites calabricae (Chevalier)

Tarbellastrea reussiana (Milne-Edwards and J.Haime) referred by ZuT fardi to Heliastrea reussiana.

These corals are thè main frame of thè reef.

Porites calabricae is different from thè other neogenic species by 3 cycles of growth, sometimes incomplete, and by thè morphology of radiai elements.

Tarbellastrea reussiana is a placoid coral, with circular chalice, tree cycles of septa, and thè lamellar col um ella due to thè fusion of septa.

Bryozoa and Melobesiae are associated with thè reef.

On this reef, where thè Porites are dominant, Melobesioids incrust thè lower necrotic parts of colonies. Often they made a film that uniformly covers thè corals. The gap between colonies of Tarbellastrea are often filled by melobesid; Bryozoans have a rule similar that thè calcareous algae.

Coffey (1972, 1977) says that Briozoans reinforce thè main frame of thè reef with their encrunstation and their growth. They encrust thè block of coral colonies filling thè gaps and occupying thè intracorallian and intracolonial cavities. They are abundant on thè surface of Porites skele- tons. The contribution of this encrustation to thè generai coesion of thè coralline building is reai, but is limited. On intracorallian filling we found Brachiopods, Molluscs, Echinids, Foraminiferids, Radiolars:

Terebratula sinuosa Brocchi

Pecten benedictus Lamarck

Flabellipecten besseri Andrejozwski

Pecten subarcuatus Tournover

Flabellipecten costisulcatus Almena

Flabellipecten flabelliformis (Brocchi)

54

Boll Soc . Natur. Napoli - Voi 102 ( 1994-1995 )

Flabellipecten pasinii (Meneghini)

Flabellipecten planosulcatus (Matheron)

Flabellipecten solarium (Lamarck)

Amussium cristatum (Fontannes)

Amussium destefani (Ugolini)

Chlamys macrotis (Sowerby)

Chlamys malvinae (Dubois)

Chlamys multistriata (Poli)

Chlamys scabrella (Lamarck)

Chlamys scabriuscula (Matheron)

Ostrea lamellosa Brocchi Pycnodonta navicularis (Brocchi)

Clipeaster campanulatus acuminatus Desor Clipeaster campanulatus redii Wright Clipeaster cottreaui Lambert-Thiery Clipeaster formosus Seguenza Clipeaster insignis Seguenza Clipeaster portento sus Desmoulins Clipeaster redii Wright Heterostegina sp

We found thè same species, although with different area and strati- graphic distribution in particularly important places of thè studied area that we can correlate to “Stretti” of Cessaniti. The fauna of arenaceous formation with Clypeastrids are uniformly distribute on thè section and thè organisms are preserved in their life position. There isn' t any partic- ular orientation. The black clay level at thè base of thè sand shows many Crassostrea gryphoides with shell boring due to Clionids and Lithophaga. The clay level is in correlation to thè levels in other locations of thè area where we found also two species of Terebralia lignitarum, Terebralia sp., rare Ostracods (. Miocyprideis ) and Foraminifera ( Pyrgo depressa, Sig- moilina tennis, Spiroloculina canaliculata, and Elphidium macellum). All those fossils give indication of particular environmental conditions.

We try to obtain by diversity index (Gleason's formula: d=S-l/logN) thè degree of heterogeneity of thè population. (Fig.3).

The presence on thè upper part of Miocene of great reef made exen- tially by Porites has been noted by Chevalier (1967), St. Martin (1990), and Bosellini (1993).

The most part of thè hypothesis are based on thè Messinian Salinity Crisis, that augmented water salinity. In thè Tortonian we can spot a few builder species following thè climate cooling. It is supposed that even thè

Barbera C. et al - A miocenic reef near Catanzaro, Calabria 55

Upper Miocene scarcity of species was due to thè upwelling of cold water from thè deep sea.

Those condition allowed thè Porites to become thè mediterranean main species. The upwelling didn't brought immediatly to reef develop- ment; thè building activity of Porites was due to thè lessening number of Diatoms, that was very common during thè upwelling.

Indice di diversità

1 t

DI D2 D3 D4 D5 D6 D7 D8 D9 DIO

Località campionamento

Figure 3 - Variation of diversity index in thè different localities of thè area.

All thè above allows to point 'taxi that thè upwelling Controls thè holigotypic Porites reef building.

This reef explosion and thè algal buildings in Mediterranean at thè end of Miocene suggests an upwelling presence along with a communica- tion with thè Atlantic Ocean.

In thè studied reef there are no holigotypicity characters, since Porites and Tarbellastrea are present with 65% and 35% respectively.

Hence we can suppose that in thè Mediterranean Sea we have two separate moments of coral buiuldings: thè first (Tortonian), with a differ- enciated fauna, and thè second (Messinian), with thè presence of thè genus Porites.

Recently (1993) Bosellini recognizes in Puglia a Pre-Messinian reef with chracters analogous to thè ones we studied.

The affinities are stili with thè Atlantic Ocean, in fact in thè Middle Miocene all thè contacts with Indo-Pacific are closed.

56

Boll. Soc. Natur. Napoli - Voi. 102 ( 1994-1995 )

REFERENCES

BARBERA C.-Tavernier A., 1988. Paleontologia della successione miocenica di Vibo Valentia. Atti IV Symp. Ec. e Paleoec. Com. Bent. Sorrento 1988. Barbera C.-Tavernier A., 1989. Osservazioni paleoambientali su un banco di ostriche del Tortoniano di Capo Vaticano (Calabria, Italia). Lavori S.I.M. Atti Congr. 1987, Sorrento, p. 409-416.

Bosellini A., 1993. La scogliera miocenica di Gagliano del Capo (Penisola salentina) e il suo inquadramento stratigrafico - sequenziale. Atti Tic. Se. Terra, 36: 33-40.

Checchia-Rispoli G., 1925. Illustrazione dei Clipeastri miocenici della Calabria.

Mem. per serv. alla descr. carta geol. dTt., Roma.

Chevalier J.P., 1961. Recherche sur les Madreporaires et les formations Récifales miocènes de la Mediterranée occidentale. Mem. della Soc. Geol. di Francia. Nuova serie, 90 (93): 1-562.

Civetta I.-Cortini M.-Gasparini P., 1973. Interpretation of a discordant K-Ar age pattern (Capo Vaticano, Calabria). Earth Pian. Se. Lt., Amsterdam, 113-118. Cortese E., 1895. Descrizione geologica della Calabria. Mem. descr. carta geolog¬ ica dTtalia, IX, Roma.

Demarq G., 1989. Pectinides neogenes: proposition d’echelle biostratigra-phique pour la Meditèrranèe. Geobios. , 23: 35-39, Lione.

Demarq G., 1990. Biostratigraphic scale of Mediterranean neogene Pecti-nids. IX R.C.M.N.S. Gong., Barcellona, 141.

De Stefani C., 1987. Il terreno terziario della vallata del Mesima. Boll. Soc. Geol. It., 4: 265-273, Roma.

Fatton E., 1973. Notes et contributions de la Province biogeographique a la Population d’apres les pectinides neogenes et actuels. Centre CERPAB, Orsay, 1-185.

Nevi ani A., 1887. Contribuzione alla geologia del catanzarese. Boll. Soc. Geol. It., 4: 169-208, Roma.

Nicotera P., 1959. Rilevamento del versante settentrionale di Monte Poro (Cal¬ abria). Mem. e note Ist. Geol. Appi, 7: 1-92, Napoli.

Ogniben L., 1973. Schema geologico della Calabria in base ai dati odierni. Geol. Rom., 12: 243-585, Roma.

Saint Martin J.P., 1990. Rècifs coralliens Miocènes d'Algèrie et du Maroc. Mem.

Mas. Hist. Nat. Paris, Se. terre, 56: 229-315.

Seguenza G., 1879. Le formazioni terziarie della provincia di Reggio (Calabria).

Mem. Cl. Se. Fis. Mat. Nat. R. Acc. Lincei, 6: 1-416.

Zuffardi R., 1937. Corallari fossili del giacimento di Cerasa, presso Cessaniti (Vibo Valentia). Pubbl. Ist. Geolog. Univ., Catania.

Boll Soc . Natur . iVapo/j - Lo/. 702 (1994-1995): 57-71 57

Deinonycho sauna (Saurischia, Theropoda): a brief group synopsis.

Signore M.

Dipartimento di Paleontologia, Università degli Studi Federico II, Napoli, Italy

Key words: Cretaceous, Deinonychosauria, Dinosaurs.

Riassunto. Deinonychosauria (Saurischia, Theropoda): un breve quadro riassuntivo.

In questo lavoro si mettono in evidenza le ultime teorie sui Deinonychosauria (Saurischia, Theropoda), insieme ad un quadro riassuntivo delle caratteristiche derivate e della paleoecologia delle due famiglie che compongono questo gruppo (Dromaeosauridae e Troodontidae).

Abstract. This work explains thè most recent theories concerning thè Dei¬ nonychosauria (Saurischia, Theropoda), as well as a synopsis of thè derived cha- racters and thè paleoecology of thè two families belonging to this group (Dro¬ maeosauridae and Troodontidae).

Proposed as systematic category by Colbert and Russel in 1969, thè Deinonychosauria are composed by two families of small carnivore di¬ nosaurs: thè Dromaeosauridae (= Dromaeosaurinae Matthew & Brown 1922) and thè Troodontidae Gilmore 1924 (recently, a new family of carnivore dinosaurs has been tentatively inserted into thè group Deinony¬ chosauria: thè Therizinosauridae (Lambert, 1994), but with many reserves, that will be exposed in thè family analysis later). The developing of these three families is limited to thè Cretaceous, with some extremely interest- ing forms. The common character these two groups share is thè peculiar claw of thè second foot digit, from which comes their common english name: “sickle-clawed dinosaurs’'. In thè past, thè forms belonging to this group were placed in thè Coelurosauria Huene 1914, a mixed group of little to medium-sized theropods, now shown as a polyphyletic group. Currently, this taxon has been splitted in many taxa, leaving thè denomi-

Received 25.6.95 , accepted 10.12.95

58

Boll Soc . A ìatur. Napoli - Fo/. 102 ( 1994-1995 )

nation Coelurosauria only to some saurischian of difficult taxonomic attri- bution. It must be pointed out that this sistematic operation is not widely accepted. Interesting is thè classifìcation proposed by G. Paul (1988), which includes thè Dromaeosauridae as a subfamily in thè Family Ar- chaeopterygidae, to which belongs thè “feathered dinosaur” Ar- chaeopteryx lithographica (Fig. 1). Paul's analysis is simple, and it’s based

Figure 1 - A possible reconstruction of Archaeopteryx (after Lambert, 1994).

on some shared characters by Archaeopteryx and Dromaeosauridae, among which there are: thè forelimb structure (chief of all thè humerus- ulna/radius articulation), thè sternal structure and thè hindlimb. Paul claims that a structure very dose to thè sickle claw of dromaeosaurids can be recognized in thè second foot digit of Archaeopteryx . The phylogenetic explaination which follows is very interesting indeed: thè Jurassic Ar- chaeopterygidae should have in an earlier time detached themselves from thè phyletic line that carries on to thè birds ( Avialae according to Gauthier 1986), and would have developed a sickle-shaped structure on thè foot claws to aid themselves while climbing trees. From these Jurassic froms would have been evolved thè Cretaceous dromaeosaurids, that would have developed thè climbing claw further, making it an efficaceous hunting weapon. Hence, thè possibility of keeping thè arms tight to thè body would have been inherited from thè Jurassic archaeopterygids, as well as many other skeletal features. Troodontids would have been derived from thè dromaeosaurids, and would belong to a second radiation of protobirds (Paul, 1988), more evoluted and more lightly-built than their precursors.

59

Signore M. - Deinonychosauria (Saurischìa, Theropoda)

Dromaeosauridae

The family Dromaeosauridae is diffused in all thè Cretaceous, from Aptian to Maestrichtian, in North America, Asia and Europe (Fig. 2). It includes some genera of very interesting predatory dinosaurs. One of thè best known is Deinonychus antirrhopus (Fig. 3), from Aptian-Albian of

Figure 2 - The geographic distribution of Dromaeosauridae.

Figure 3 - The reconstruction of Deinonychus antirrhopus by R. T. Bakker (after

Ostrom, 1969).

North America. Another known form among Dromaeosaurids is Ve- lociraptor mongoliensis (Fig. 4), from thè late Santonian or thè early

60

Boll Soc . Natur. Napoli - Voi 102 ( 1994-1995 )

Campanian, found in Mongolia and China. Gauthier (1986) States that this group is dose to birds. Paul, as before noted, suggests (1988) that thè Dromaeosaurids may be thè direct, non-flyer descendants of thè Jurassic Archaeopterygids.

Figure 4 - A running V elociraptor mongoliensis (after Lambert, 1994, modified).

Among thè characters that mark thè family Dromaeosauridae, a com¬ plete list of which is beyond thè goals of this work, it must especially be noted thè remarkable modification of thè hindlimbs, that bear a very modified foot, compared to thè standard theropod tridactyl condition, since it becomes didactyl, with thè II digit hearing thè sickle-shaped claw (Fig. 5), which makes thè dromaeosaurs so famous. The elongated zygapo-

Figure 5 - The typical dromaeosaurid foot (in this case, thè Deinonychus foot). Note thè fearsome sickle claw (after Ostrom, 1969).

physes suggest that these dinosaurs used thè tail as a balancing structure, hence allowing hypothesis about thè lifestyle of thè dromaeosaurs, as described further. Even thè pelvic girdle is unique to this group.

Signore M. - Deinonychosauria (Saurischia, Theropoda)

61

The family is divided into two subfamilies, thè Dromaeosaurinae and thè Velociraptorinae, and this division is supported by clear differences in thè teeth shapes (Currie et al., 1990).

A detailed anatomical descritpion can be found in J. Ostrom’s work (1969) on thè Deinonychus antirrhopus. Here it’s pointed out one remark- able fact; thè “posterior trocanther” on thè femur (which, by thè way, is known only from Deinonychus and Velociraptor), that should be an attach- ing place for a typical dromaeosaurid muscle, related to their unusual mode of attack (Norman, 1985; Ostrom, 1990).

The dromaeosaurids must have been predators that hunted in large groups in thè Cretaceous plains. Their attack mode was different from thè way of attack used by thè great theropods (e.g., thè tyrannosaurids), that cut off great pieces of flesh with powerful bites, so debilitating thè prey for shock and copious bleeding. The presence of a rigid tail (Fig. 6), a “poste¬ rior trocanther", and their famous sickle claw let us hypothize that thè

Figure 6 - The rigid encasement of Dromaeosaurs tail, an ideal aid to hunt (after

Ostrom, 1990).

dromaeosaurids used their claws in thè same way thè other great theropods used their jaws, i.e. to critically wound (and hence to debilitate) their preys (Paul, 1988; Ostrom, 1990). The stiff tail would have been used as a stabilizer at thè moment of thè attack, and thè posterior trocanther were to be an attaching point for a particular muscle, used to move thè hindlimbs to kick, and in thè same time to avoid using thè caudifemoralis in thè retraction of thè ìimb, that would have moved thè tail, thus umbal- ancing thè animai. The brain volume of thè dromaeosaurids is relatively elevate, but only if compared to a “dinosaurian ” standard, and cannot compete with a mammalian brain. Bakker (1986), Paul (1988) and other Authors advanced thè hypothesis that thè dromaeosaurs, as well as other

62

Boll Soc . Natur. Napoli - Fo/. 102

-1

little carnivores among thè saurischians, would have feathers (Fig. 7). This is possible, since their methabolic level should be rather high. Should be this thè case, we can imagine thè dromaeosaurs as great running birds, perhaps having brilliant colours and complex behavioural patterns.

TwW-'*

Figure 7 - Gregory Paul's reconstruction of two feathered Deinonychosauria: a Velociraptor (on thè left) and a pair of Troodon (on thè right), quarreling about

a prey (after Paul, 1988).

Evolution. According to J. Ostrom (1990), thè evolutive lineage that brings to Dromaeosaurs should be traced from a stock that includes forms as Coelurus and Ornitholestes. In thè former can be seen thè asymmetry of thè radiale, a character of thè Deinonychosauria, while in Ornitholestes has developed a second foot digit with a large claw, thus precurring thè didactyl condition of thè Dromaeosaurs. There aren’t instead precise evi- dences about when thè split between Dromaeosauridae and Troodontidae occurred, but it is known that thè two groups become distinct in Asia at thè end of thè Santonian, and in North America about in thè middle Campanian. The most recent species are derived from a basai stock, that seems to be composed (until now) by Deinonychus in North America and by Velociraptor in Asia. Since Paul (1988) proposed, as stated below, thè synonymy of thè two genera (along with thè S aur ornitholestes) , if his hypothesis should be right, then we’ll have to consider a wider distribution

Signore M. - Deinonychosauria (Saurìschìa, Theropoda)

63

of thè sole genus Velociraptor, that hence would be thè basis for thè other genera, and would have his own evolution from thè species V. antirrho- pus, to V. mongoliensis and fìnally to V. langstoni (thè latter was a small dromaeosaurid from thè late Campanian of Alberta, Canada). For what concerns thè phylogenetic relationships with thè archeopterygids, even Ostrom (1990) and other Authors agree to state a tight relationship be- tween dromaeosaurs and archeopteryigids, albeit they express an opposite idea respect to that of Paul (1988), i.e. that Jurassic dromeosaurs should be thè ancestors or thè archeopterigids and birds (Ostrom 1990). There should be a confirmation to this theory only with then findings of Jurassic Dromeosaurs, but until now it seems that this group knew a wide expan- tion only at thè end of Cretaceous, thus making a great obstacle to thè formerly exposed idea, while confirming Paul’s hypothesis.

Paleoecology. The sedimentological conditions of thè strata from which either Deinonychus and Dromaeosaurus come, in North America, mark alluvial plains of large rivers. The Mongolian deposits where Ve¬ lociraptor has been found indicate a more arid area.

Dromeosaurs have been found even in Europe. Until now, they have been described from France, in thè Aude area (Le Loeuff & Buffe ttaut, 1993).

Fossil evidences help thè paleontologists to state with an adequate certainty what were thè preys of these fast hunters. In thè case ot Deinony¬ chus, there have been findings of thè teeth of this dromaeosaurid among thè bones of a possible iguanodontid, Tenontosaurus tilleti Ostrom 1970, as well as at last four almost complete individuai around thè skeleton of one of these great iguanodontids. This could mean that Tenontosaurus should have been thè typical prey for Deinonychus, and thè presence of four (or maybe Uve) Deinonychus around only one Tenontosaurus could mean that these dromaeosaurids were pack hunters (Norman, 1985). In Mongolia were found a complete skeleton of Velociraptor litterarly “clasped” with a Protoceratops andrewsi Granger & Gregory 1923 (a small protoceratopid), and this should unequivocally mark one of thè typical preys of Velociraptor (Norman, 1985).

Dromaeosaurian stride was bipedal, when they ran as well as when they walked, and was by alternate steps, by any mean not jumping; but in thè fossil record there aren't trackways surely ascribed to Dromaeosaurs.

The Dromaeosaurs' forelimb were notably strong and with a good grasp, but thè attack against thè preys were conducted with both thè

64

Boll Soc . Natur, Napoli - Voi 102 ( 1994-1995 )

hindlimbs, and this hypothesis is supported by thè presence of a long, rigid tali.

Without doubt, thè Dromaeosaurids, along with their “cousins” Troodontids, are among thè most fascinating and strange predators that ever roamed thè Earth.

Troodontidae

The other family in thè group Deinonychosauria is constituted by thè Troodontidae Gilmore 1924. This family is one of thè less represented in thè fossil record, since it includes only three certain genera (and a few probaie ones), instituted on fragmentary remains, and many nomina dubia (see below in thè systematic appendix). Nevertheless, thè findings of thè Chinese-Canadian expedition in Mongolia show more skeletal evidences belonging to this family, although they are stili in study. All thè Troodon¬ tids come from Cretaceous deposits of thè Northern hemisphere (Fig. 8),

Figure 8 - The geographic distribution of Troodontidae.

and they are distributed from thè Aptian-Albian in Mongolia until thè North American Maestrichtian. In thè last years these theropods were known as Saurornithoididae Barsbold 1974, but in 1987 Currie proved that thè correct name was Troodontidae, instituted by Gilmore on some

65

Signore M. - Deinonychosauria (Saurischia, Theropoda)

m

teeth attributed to thè Pachycephalosauria (Marginocephalia, Ornithis- chia), that instead turned out to be very dose to thè Sauro rnithoididae teeth.

Until thè recent pubblication of some results of thè Chinese-Canadian expedition in Mongolia, thè fragmentary remains of these beings didn’t allowed precise reconstruction (e. g., thè vertebral column was not known).But thè new findings have confirmed that thè troodontids were small theropods, fast and lightly built, with slender limbs,and blessed with thè most developed brain among thè Dinosaurs. From this character, Russel & Sèguin (1982) elaborated a curious creature, thè “dinosauroid”, i. e. a very evolved descendant of thè troodontids, thè “archosaurian hominid”...

The anatomical descriptions for these theropods are stili in progress. In thè skull there are many distinctive features, among which there are thè great number of teeth (25 in thè maxillary, and almost 35 in thè dentary), that are very small and hooked and also very dose to each other, and thè lateral depression on thè side of thè braincase that contained thè middle ear cavity.The post cranial skeleton shows a slender and light constitution, with long tridactyl forelimbs, and thè hindlimbs that show a raptorial claw, similar to thè one observed in thè Dromaeosauridae, but smaller and maybe no longer with thè same function it has in their larger cousins. According to Currie et al. (1990), thè teeth are indicative, at least to a generic level. A complete list of characters can be found in Currie and Dong (1993).

Evolution. The systematic position of thè troodontids among thè theropods is stili not clear. The forelimb structure seems to be a primitive character, common in all thè small theropods (except in thè Ornithomim- idae, which are a derived group), while a crescent-shaped carpai bone joins thè troodontids to thè Maniraptora Gauthier 1986, more precisely to thè Deinonychosauria (Barsbold 1974; Osmolska 1982). Currie (1985, 1987) suggests that they are closer to thè birds for some characters, among of which thè pneumatized bones, but this hypothesis is seriously disputed (Barsbold et al.; also Osmolska 1976).

Currently, it seems more correct to have thè Troodontidae and thè Dromaeosauridae grouped together in thè same group due to some shared characters (among of which thè sickle -claws and thè sagittal ridge).

Troodontids make their appearance for thè first time in thè Lower Cretaceous of Asia, and do not modify during their evolution some basai

66

Boll Soc . Natur. Napoli » Voi 102 ( 1994-1995 )

characters, showing as it seems a shortening of thè II pedal falanx of (Barsbold & Osmolska, 1990).

Paleoecology. Troodontids were small carnivores, with a very acute stereoscopie vision and a well developed brain (Fig. 9), thus making

Figure 9 - Troodon formosus, drawn in scale with a house cat (after Lambert,

1994).

dangerous predators, maybe active even by night. Moreover, thè high activity levels they show make them very suitable for omeothermy, more than any other dinosaur. To discuss thè matter of omeothermy in di¬ nosauro is a very difficult question, and is far beyond thè scopes of this paper to dispute this subject, hence thè hypothesis is just reported for completeness. The sensory imput must have been impressive: along with their powerful sight, thè Troodontids show a very keen hearing, as testified by thè great middle-ear cavity and thè periotic sinum, although it seems that thè smeli was not so developed. The typical preys for these fast predators were with any probability lizards, mammals, dinosaur hatch- lings and, maybe, inseets. The "night mammal hunter” hypothesis is very fascinating, but generally thè paleontologists opt for a daytime activity cycle.

The sickle shaped claw was possibly used in a different way than in thè Dromaeosaurs. This statement is supported by thè structural differences between thè two families in thè hindlimbs (with a consequent less power in thè “kick” of thè Troodontids), and by thè fact that in Borogovia thè claw is no longer hooked, but straight, and that seems to be a trend for thè

Signore M. Deinonychosauria ( Saurtsckìa , Theropoda)

67

Troodontids, thus they rarely relied on thè claw in their attacks, possibly using their speed, and maybe accounting on their hunting tactics.

The rare and fragile remains of thè Troodontids are found in sedi- ments that show an arid climate, as in thè case of Velociraptor mongolien- sis, that was contemporary of Saurornithoides mongoliensis , or in fluvial deposits, marked by thè presence of great herbivores ( Saurolophus Brown 1912) and carnivores ( Tarbosaurus bataar Maleev 1955, bui. some Authors refer to it as a subgenus of Tyrannosaurus , e. g. Paul 1983), for what about thè Asian Troodontids» In North. America thè only known Troodontid was Troodon formosus, that lived in a humid and forest-rich environment. In any case, thè Troodontids represent only a small part of thè fauna in which they lived, and should have been rare indeed.

The only sure presence of Troodontids in Europe comes fi oro thè Maestrichtian deposits in Romania, but thè remains (named Bradycneme draculae and Heptasteornis andrewsi) are poor indeed, and are also re • worked, so nothing can be told about thè environment in which they lived (Paul, 1938; Barsbold & Osmolska, 1990). Kecently, two possible Troodon¬ tid dinosaurs,one named Lìsboasaurus , from Portugal (Milner & Evans 1991), and thè other recognized in a sacrum from thè Wealden of thè Isle of Wight (Howse & Milner 1993), have been also described.

A last (but not least) series of data comes from thè microscopie analysis of thè Troodon bones. It seems from this analysis that Troodon reached its adult size (about 50 kg) in 3-5 years, and then grew no longer, at least not significantly. The bone growth is made up of three phases: thè fìrst, very rapid, with a fib rou s • lame 1 1 ar bone; thè second, more moderate, with lamellar-zonate bone; and thè third, at thè end of thè growth, was a slow depositìon of a lamellar avascular subperiostal bone. Furthermore, thè structure of bone, with thè three stages of growth, along with thè high EQ (Encefalization Quotient) level and thè cursorial adaption confirm thè hypothesis of thè high activity levels reached by thè Troodontids, and maybe should be thè case of accepting thè idea of an omethermy, at last a partial one (V arriccino 1993). By what means did they reached it, is thè subject of a very interesting dissertation (more on McNeill Alexander, 1989).

APPENBIX ONE: SYSTEMATICS

Here is reported thè most recent list of genera attributed to Dromaeosauridae and Troodontidae (mainly from Weishampel et ah, 1990, modifìed).

68

Boll Soc . Natur . Napoli - Voi . 102 ( 1994-1995 )

Dromaeosauridae Matthew & Brown 1922 Adasaurus mongoliensis Barsbold 1983 Deinonychus antirrhopus Ostrom 1969 Dromaeosaurus albertensis Matthew & Brown 1922 Hulsanpes perlei Osmolska 1982 Saurornitholestes langstoni Sues 1978 Velociraptor mongoliensis Osborn 1924

Nomina dubia dromaeosauridae

Koreanosaurus not described Phaedrolosaurus ilikensis Dong 1973

To these genera, some add thè genus Ghiro stenotes , with thè species C. pergracilis Gilmore 1924, albeit some other report it as a Caenagnathid, hence an Oviraptorosaur (Currie & Russell, 1988; Lambert, 1994). Re- cently has been found a large Dromaeosaurid, described in 1993 (Kirkland et al., 1993) and named Utahraptor. This giant reached thè size of about 8 meters.

Note also that Paul (1988) proposed to join thè three froms Deinony¬ chus antirrhopus, Saurornitholestes langstoni, and Velociraptor mon¬ goliensis under thè sole genus Velociraptor as three different species. Of course, Ostrom (as well as other paleontologists) doesn’t agree. Anyway, until new finds, we consider them three separate genera.

Troodontidae Gilmore 1924 (= Saurornithoididae Barsbold 1974) Troodon formosus Leidy 1856 (includes Polyodontosaurus grandis Gilmore 1924, Stenonychosaurus inequalis Sternberg 1932, and Pectinodon bakkeri Carpenter 1982)

Saur ornithoides mongoliensis Osborn 1924 Saurornithoides junior Barsbold 1974 Lisboasaurus estesi Seiffert 1973 Borogovia gracilicrus Osmolska 1987 Sinornithoides youngi Currie & Dong 1993

Nomina dubia troodontidae

Bradycneme draculae Harrison et Walker 1975 (some consider it a Droameosaurid)

Signore M. - Deinonychosauria (Saurischia, Theropoda)

69

Elopteryx nopcsai Andrews 1913 Heptasteornis andrewsi Harrison et Walker 1975 Paronychodon lacustris Cope 1876 Tripriodon caperatus Marsh 1889

Zapsalis abradens Cope 1876 (possibly another name for Parony¬ chodon)

The many nomina dubia are due to thè paucity of thè fìndings (all are based upon teeth or a few broken bones), and to these it must be added another unnamed Troodontid (Barsbold et ah, 1987).

The systematic situation of Troodontids is very confused, mainly for thè scarcity of remains, and only Currie's work (1987) made so me order, by joining three genera in just one, and using thè old family name instead of thè one proposed by Barsbold. The work on thè teeth (Currie et al., 1990) is also very important, and could give more light to an otherwise dark, but fascinating, group.

APPENDIX TWO: THE THERIZINOSAURUS

The Therizino saur us has been described in thè appendix for two rea sons: first, it is not well known, as will be shown, and second, it is considered belonging to thè Deinonychosauria with many doubts, since it is generally included in thè “problematic” Coelurosauria (Norman, 1990), and some claims this may be a Segnosaurid dinosaur (Lambert, 1994). Anyway, thè genus includes only one species, thè Therizino saur us cheioni - formis Maalev 1954, known only from few bo nes of a forelimb, formerly considered to belong to a tortoise. Since it has been found in thè same area of other ill-documented dinsaurs (e.g. Deinocheirus mirificus Osmol- ska et Roniewicz 1970), it cannot be justified to form a family, but surely thè characters that can be observed in thè few known bones separate it from other odd dinosaurs of Mongolia. If thè typical body proportions for a Deinonychosaur are respected for this genus, thè Therizino saurus would have been over 10 meters long. A truly monster in thè group, indeed!

The Therizino saurus comes from thè Upper Cretaceous of Mongolia.

Many questions are stili open about thè Therizinosaurids, and only further finds will proove that any theory may be right. In some re cent papers (Russel and Dong, 1993), thè position of Therizinosau rids are further examinated, resulting in dose relationships with “segnosaurs”. According to thè two Authors, thè Therizinosauridae, along with thè

70

Boll Soc . Natur. Napoli - Voi. 102 (1994-1995)

Troodontidae, should be moved into thè group of Ovi raptorosauria, while thè Dromaeosauridae go among thè Carnosauria (along with Tyran- nosauridae and other great carnivores). Since this subjects is not explored in this paper, it must be referred to thè paper of Russel and Dong (1993) for further informations.

AC KNO WLEDGEMENTS

I would like to thank prof. L. Barbera of thè Department of Paleon- tology of Naples for her precious advice, and prof. A. Kotsakis of thè Department of Earth Sciences of University of Roma III, Rome for advice and thè criticai revision of thè manuscript.

REFERENCES

Barsbold R., 1974. Saurornithoididae, a new family of small thero pod dinosaurs from centrai Asia and North America. Palaeont. Polonica 30: 5-22.

Barsbold R., Osmolska H., & Kurzanov S. M., 1987. On a new troodon tid (Dino¬ sauria, Theropoda) from thè Early Cretaceous of Mongolia. Acta Palaeontol. Polonica 33: 49-52.

Barsbold R., & Osmolska H., 1990. Troodontidae. In Weishampel D, Osmolska H., Dodson P. (Eds.), The Dinosauria. University of California Press, Berkeley. Colbert E.H., Russell D. A., 1969. The small Cretaceous dinosaur Dromaeosaurus . Am Mus. Novitates 2380: 1-49.

Currie P.J., 1985. Cranial anatomy of Stenonychosaurus inequalis (Saurischia, Theropoda) and its hearing on thè origin of birds. Can. J . Earth Sci. 22: 1643-1658.

Currie P.J., 1987. Bird-like characteristics of thè jaws and teeth of troodontid theropods (Dinosauria, Saurischia). J. Veri. Paleontol. , 7: 72-81.

Currie P.J., 1995. New information on thè anatomy and relation ships of Dromaeo¬ saurus alhertensis (Dinosauria: Theropoda). J. Vert. Paleontol., 15 (3): 576- 591.

Currie P.J., & Russell D.A., 1988. Osteology and relationships of Chirostenotes pergracilis (Saurischia, Theropoda) from thè Judith River (Oldman) Forma- tion of Alberta (Canada). Can. Jour. Earth Sci., 25: 972-986.

Currie P.J., Keith Rigby Jr. J., & Sloan R.E., 1990. Theropod teeth from thè Judith River Formation of Southern Alberta, Canada. In Carpenter, K., e Currie, P.J. (Eds.): Dinosaur Systematics: Ap proaches and Perspectives, Cambridge U. Press.

Gauthier J., 1986. Saurischian monophyly and thè origin of birds. In: Padian, K. (Ed.), The Origin of Birds and thè Evolution of Flight. Mem. Calif. Acad. Sci. 8: 1-55.

Signore A4. - Demonychosauria (Saurisckìa, Theropoda)

71

Kirklamo J.I,D Burge, & R. Gaston, 1993. A large droxxiaeosaur (Theropoda) from thè Lower Cretaceous of eastem Utah. Hunteria 2: (10): 1-16.

Lambert IL. 1994. The Encyclopedia of Dinosaurs. Bloomsbury Books, London, in association with thè British Museum (Naturai History), London.

LeLoeuff J . Buffettaut E., Mechin P.f & Mechxn-Salessy A., 1992. The fìrst record of dromaeosaurid dinosaurs (Saurischia, Theropoda) in thè Maastrichtian of southern Europe: palaeobiogeographical im plications. Bull Soc. géol Frane e 163: 337-343.

Le Loeuff J., & Buffettaut E., 1993. Gli ultimi dinosauri della Francia del sud. Le Scienze Quaderni n° 76, (1994) “I Dinosauri”.

Norman D.B., 1985. The Illustrated Encyclopedia of Dinosaurs. Crescent, New York.

Norman D.B., 1990. Problematic Theropoda: "Coelurosaurs”. In Wexshampel D , Osmolska H.» Dodson P. (Ed.), The Dinosauria. University of California Press, Berkeley.

Me Nexll Alexander, R., 1989. Dynamics of dinosaurs and other extinct giants. Columbia U. P., New York.

Osmólska H , 1976. New light on skull anatomy and systematic posi tion of Oviraptor. Nature 262: 683-684,

Osmólska H , 1982. Hulsanpes perlei n. g. n. sp. (Oeinonychosauria, Saurischia, Dinosauria) from thè Upper Cretaceous Barun Goyot For mation of Mongolia. N. Jb. GeoL Palàontol Mh. 1982: 440-448.

Ostrom J.H., 1969. Osteoìogy of Deinonychus antirrhopus , an Unusual Theropod from thè Lower Cretaceous of Montana. Peabody Mus. Nat. Hist. Bull. 30: 1-165.

Ostrom J.H., 1974. The pectoral girdle and forelimb funcii.cn of Deinonychus (Reptilia: Saurischia): a correction. Peabody Museum Postilla, 165.

Ostrom J.H., 1990. Dromaeosauridae. In Weishampel D Osmolska H . Dodson P. (Ed.), The Dinosauria. 1990 University of California Press, Berkeley.

Paul G.; 1988. Predatory dinosaurs of thè world. Simon and Schuster, New York.

Russell D A, 1969. A new specimen of Stenonychosaurus from thè Oldman For¬ ma non. (Cretaceous) of Alberta. Can. J. Earth Sci 6: 592-612.

Russel O.A., & Dong, 1993. A nearly complete skeleton of a new troodontid dìnosaur from thè Early Cretaceous of thè Ordos Basìn, Inner Mongolia, Peopie's Republic of China. Can. J. Earth Sci. , 30: 2163-2173.

Russel D A , & Dong, 1993. The affinities of a new theropod from thè Alxa Desert, Inner Mongolia, Pcople s Republic of China. Can . J. Earth Sci, 30: 2107-2127.

Russell D.A., & Séguin R. 1982. Reconstruction of thè small Cretaceous theropod Stenonychosaurus inequalis and a hypothetical di nosauroid. Syllogeus 37: 1-43.

Sues H.D., 1977. The skull of V elocìraptor mongoliensis , a small theropod dinosaur from Mongolia. Palàontologische Zeitschrift , 51: 173-184.

Varricchio DJ., 1993. Bone microstracture of thè Upper Cretaceous theropod dinosaur Troodon formosus. J. Vert. Paleont. 13 (1): 99-104.

.

■

Boll Soc. Natur . Napoli - Voi 102 (1994-1995): 73-7 9

73

Study on Glass components from Soils of Phlegrean Fields. First Note. The glasses in thè sands of Monte Spina (Agnano)

Pugliano M.L., Murolo* M. and Buondonno C.

Dipartimento di Scienze Chimico- Agrarie, Facoltà di Agraria, Università degli Studi di Napoli “Federico II”

Key words: volcanic glass; soil

Riassunto. Studi sui componenti vetrosi dei suoli dei Campi Flegrei: i vetri delle sabbie del Monte Spina (Agnano).

È stata esaminata la componente vetrosa presente nelle sabbie separate da campioni di suoli sviluppati sulle vulcaniti di Monte Spina (Agnano). È stata apprezzata la percentuale di vetri (Volcanic Glass Test) dei quali sono stati osser¬ vati e descritti i caratteri morfologici e costituzionali (microscopia ottica e diffrat- tometria ai raggi X). Lo studio, finalizzato ad integrare la valutazione dell'andicità dei suoli flegrei, è stato completato con analisi chimiche dei campioni di suolo.

Abstract. We examined thè glass components present in thè sands of Monte Spina (Agnano).

We estimated thè glass content (Volcanic Glass Test), and its morphological and constitutional characters were observed and described (optical observation and X-ray diffraction analyses).

The study, which aims to complete thè evaluation of andic properties of Phlegrean soils, also consisted of chemical analyses of soil samples.

Introduction

Aceording to thè most updated criteria of soil classification, defined by Soil Taxonomy (Soil Survey Staff, 1992), thè primary glass content of volcanic glass sands, with reference to other soil characteristics, is one of thè andic features, and is thus a parameter for thè classifying thè order of Andisols and for their differentiation from other taxonomic orders.

* M. Murolo performed chemical analyses on samples, while M.L. Pugliano performed thè mineralogical study on sands.

Receìved 12.5.95 , accepted 12.12.95

74

Boll Soc . Natur. Napoli - Voi 102 ( 1994-1995 )

In thè Phlegrean Fields thè outcrops consist mainly of pyroclastic materials, deposited over a very long period, from 47,000 years B.P. to 1538 (Di Girolamo et ah, 1984; Rosi & Sbrana, 1987). This suggests that thè pedogenesis gives rise to soils with features of Andisols, Inceptisols and Entisols.

A contribution to thè interpreting properties and capabilities of Phle¬ grean soils, and also to defining pedogenesis evolution and clarifying classification, can be provided by a consistent definition of thè content, nature and condition of glass particles in thè sand fraction. In fact, thè development stage of glass components contributes to interpreting thè state of alteration of thè soil profile. In thè Phlegrean Fields thè pedolog¬ ica! environment has also been affected by thè rejuvenation of thè soil on thè surface, through successive eruptive events, which makes more com- plex thè interpretation of current pedogenetic processes.

The purpose of this paper was to study thè sandy glass components from thè Phlegrean soils and to examine thè relationships between their nature and soil classification.

The results of investigation of one representative soil of thè Monte Spina pedologica! environment are reported here. The Monte Spina is one of thè four units that constitute thè Agnano volcanic complex, located in thè Phlegrean Fields’ south-eastern area. The products of thè Agnano volcano have been estimated to be younger than 7000 years B.P. (Lirer et ah, 1987; Rittmann, 1950) (Fig. 1).

Materials and methods

The soil samples for laboratory analysis were collected from horizons of thè selected soil profile.

Samples of thè soil were air-dried and thè sand fraction with particle size 2.0 - 0.02 mm was separated by sieving. H202 and acetic buffer pretreatments were conducted to remove organic matter and coatings and patinas respectively (SISS, 1980).

The sands were examined with stereo-binocular for thè observation of glass component.

The Volcanic Glass Test (V.G.T.) (Icomand, 1984) was carried out to estimate thè contents of glass particles. The percentage ratio of glass to crystal components that are involved in thè sand formation is a distinguishing feature of thè Andisols. When combined with thè contents

Pugliano M.L . et ai - Study on Gtass- componènts of Phlegrgan Fieldsf 75

of Ài + 1/2 Fe oxalate exlractable, thè ratio reaches thè minimum value estimateci by thè classifìcation (Soil Survey Staff, 1992).

The sands were also examined in thin sections with polarising mi- croscopy.

study area

After crushing, some selected glasses were examined by x-ray dii. frac tion procedures, using a Seifert 400 diffractometer, with Ni fìltered CuKot radìation, obtained at 40 kv and 30 mA.

To estimate thè andic properties of thè soil (Soil Survey Staff, 1992), on < 2 rum fraction of samples obtained by sieving after peroxidation, Al

76

Boll. Soc. Natur. Napoli - Voi. 102 (1994-1995)

+ 1/2 Fe oxalate extractable (Schwertmann, 1964) and thè phosphate absorption (Blakemore et al., 1987) were determined. Such analyses provide an indication as to thè content of newly-formed glass material.

Results and discussion

Mineralogical observations carried out on sand fractions showed that thè samples consist of five kinds of grains.

A first kind of grain has a polyhedral or lengthened shape, with more or less rounded corners, a pale grey-yellowish colour, a vitreous look with inclusions of k-feldspar, plagioclase, biotite, pyroxene and magnetite phe- nocrists. They are typical pumice, highly vesicular; in some of them thè vesicles have a canal shape because of their pyroclastic flow origin. (Fig. 2).

A second type of grain has a rounded shape, a dull white colour, and a vitreous nature, with dark red inclusions. From thè results of x-ray diffraction analysis it emerges that thè latter consist of mainly amorphous material and only subordinately of pyroxenes. (Fig. 3).

A third kind of grain has a light black colour and a polyhedral shape, with sharp corners and conchoid fracture; thè mass is vitreous or with some rare phenocrists: these particles are typical obsidian. On thè surface there are small size vesicles full of yellowish material that is amorphous.

A fourth kind has a polyhedral shape with rounded corners, a pale grey and white colour with little black inclusions; through polarising microscopy analysis, it emerges that thè pale grey mass consists of glass, thè white mass consists of sanidine crystals and thè black inclusions of magnetite microphenocrysts.

A fifth kind of grain has a dark grey to dark brown colour, a polyhedral shape with sharp corners and it is highly vesicular: these particles are typical scoria.

From thè results of V.G.T., thè volcanic glass content was 57%, 79% and 89% in thè three horizons respectively, and it also increases along thè profile. The glass components consisted mainly of grains with a recognis- able morphology, such as that of pumice, scoria and obsidian. They ap- peared to consist of primary glass with little pedogenetic alteration. The red or yellow surface inclusions, made up of minerai amorphous to x-ray material, represent an alteration product of thè primary glass. The obser¬ vations performed show that thè alteration involved thè glass grain only in part. This suggests that thè desegregation and hydrolysis process of thè

Figure 2 - Highly vesicular grain with inclusion of pyroxene phenocrist.

Figure 3 - Glass grain with dark red amorphous inclusions.

Boll. Soc. Natur. Napoli - Voi. 102 (1994-1995)

78

alluminosilicate glass materials is in its initial phase, which is indicative of a short pedogenetic evolution. This assertion is confirmed by preserva- tion conditions of thè primary glasses.

The results of Chemical analysis showed that thè Al + 1/2 Fe oxalate extractable values and thè phosphate absorption decrease from 0.3% to 0.1%, and from 21% to 19%, respectively, along thè profile. These data express thè primary glass alteration process which proceeds from thè surface into deep layers. However, such alteration was not able to give advanced andic features for which at least a 0.4% Al + 1/2 Fe oxalate contents and 25% phosphate absorption are required (Soil Survey Staff, 1992).

The results of mineralogical observations are consistent with those of Chemical analysis and help us to understand more clearly thè soil evolu¬ tion stage on thè Monte Spina formation. The weak alteration of primary glasses, thè low contents of secondary glasses and thè consequent low phosphate absorption values indicate a weak development of andic prop- erties. This outcome excludes thè classification of thè Andisol order. Therefore thè andic characteristic is shown to be in a lower taxonomic class, which is thè vitrandic subgroup of thè Inceptisol order.

Acknowledgements.

The authors are grateful to Prof. Enrico Franco (Dipartimento di Scienza della Terra, Università degli Studi di Napoli, Federico II) for kindly performing thè thin sections.

REFERENCES

Blakemore L.C., Searle P.L. and Daly B.K., 1987. Methods for chemical analysis of soils. N.Z. Soil Bur. Sci. Rep. 10A. Rev. Ed.: N.Z. Soil Bur.

Di Girolamo P., Ghiara M.R., Lirer L., Munno R., Rolandi G. and Stanzione D., 1984. Vulcanologia e petrologia dei Campi Flegrei. Boll. Soc. Geol. It., 103: 349-413.

Dixon J.B., Weed S.B. (eds), 1989. Minerals in soil environments. Soil Science Society of America, Madison, Wisconsin.

Icomand (International Committee on thè classification of thè Andisols), 1984.

Circular letter n. 6.c/ - Soil Bureau D.S.I.R., Lower Hutt, New Zealand. Lirer L., Pescatore T.S. 1987. Geologia delle aree di Monteruscello e del centro storico di Pozzuoli. Progetto Pozzuoli, quaderni di documentazione n° 1.

Pugliano M.L . et ai - Study on Glass components of Phlegrean Fields 79

Macco G.B., 1983.: Il microscopio polarizzatore e le sue applicazioni in petrografia,

Edizioni Kappa, Roma.

Mackenzie W.S., Guilford C., 1985. Atlante dei minerali costituenti le rocce in sezione sottile, Zanichelli.

Rittmann A., 1950. Sintesi geologica dei Campi Flegrei. Boll Soc . Geol. It ., 69: 1 17=362.

Rosi M., Sbrana A., 1987. Phlegrean Fields. CNR, Quaderni della ricerca scientif¬ ica, 114, voi. 9, Roma.

Shwertmann U., 1964. The differentiation of iron oxide in soils by a Photochemical extraction with acid ammonium oxalate. Zeit. Pflanz . Dueng . Bodek., 105: 194-201.

SISS., 1980. Metodi normalizzati di analisi del suolo. Edagricole, Bologna.

Soil Survey Staff, 1992. Keys to Soil Taxonomy. 5th ed. Soil Manage. Support Serv. Tech. Monogr. 19. Blacksburg, Virginia PocaHontas Press, Ine.

'

Boll Soc . Natur. Napoli - Voi 102 ( 1994-1995 ): SI -90 81

Analytical procedure to study animai communities structure’s in fragmented environments

Domenico Fulgione, Patrizia Mirabella and Mario Milone

Dipartimento di Zoologia, Università degli Studi Federico II, via Mezzocannone 8, 80126 Napoli, Italy

Key words: multivariate analysis - animai communities - patches - birds.

Abstract. This paper suggests a procedure to estabilish a relation between thè patches and their animai communities. The method is based on a doublé ordina- tion diagram through a multivariate analysis. The first is structured on thè zooce- notic component and thè second is represented by all thè landscape’s elements that characterize thè site. The variations between thè positions of thè sites in thè two ordination diagrams fìx some informations of thè importance of thè patche. In this work is exemplified one application using twelve sites of thè campanian region and their birds communities.

This procedure could offer reliable results if it is applied to sites with similar environmental typologies, reducing, for those, thè variabilities examined.

Riassunto. Procedura analitica per lo studio della struttura delle comu¬ nità animali in ambienti a mosaico.

Lo studio propone una procedura per valutare l'importanza relativa delle tessere di un mosaico ambientale in funzione della comunità animale presente. Essa si basa su una doppia classificazione dei siti, scelti come campioni rappresen¬ tativi della relazione tessera-fauna. La prima relazione di somiglianza è data dall’informazione zoocenotica e utilizza l’Analisi delle Componenti Principali’, la seconda si basa sulle informazione provenienti dai parametri ambientali mediante Cluster Analysis. Il confronto dei due diagrammi porterà a formulare ipotesi sull'importanza di singole componenti ambientali nel determinare le comunità animali di ciascuna tessera. Il metodo si adatta a varie situazioni di studio, nel presente elaborato si espone un esempio costruito su dodici siti della regione Campania caratterizzati in base alle tipologie ambientali ed alle specie di uccelli nidificanti.

Introduction

The grasp of thè complex interaction between animai communities and other environmental factors represents one of thè major problems in thè ecological studies of thè communities.

Received 23.6.95, accepted 02.9.95

82

Boll Soc. Natur. Napoli - Voi 102 ( 1994-1995 )

In field-works, thè ecologist can’t examine separately thè environ- mental parameters. On thè contrary thè researcher studies frequently thè variations of thè animai communities in relation with thè environmental factors as a whole.

Moreover it is difficult to individuate thè single factors influencing thè communities and to evaluate their respective importance.

The analysis of thè gradients is one of thè procedures normally used to semplify thè study of such complex interaction (Terborgh, 1970).

This procedure makes possible to evaluate thè modifications of thè animai communities with reference to that environmental parameter changes more evidently then other factors.

Yet, thè analysis of thè gradient doesn't assure that is thè only one factor changing and thè choise of such factor is always subjective.

In thè latest ten years, many ecological works appeared using multi- variate analysis: with this procedure it is possible to arrange thè data, themselves by themselves.

In this way thè researcher doesn't introduce preconceived ideas about thè manner which species tend to associate between themselves or about thè manner which environmental parameters are correlated with thè distribution of thè species.

This paper suggests a procedure to estabilish a relation between thè patches, arisen by a process of landscape fragmentation, and thè animai communities that live there.

This procedure assumes that each site presents two components: thè first, thè zoocenotic component, is represented by all thè individuals of thè animai taxa ; thè second is represented by all thè landscape’s elements that characterize thè site.

Consequently, it is possible to evaluate thè similarity degree between (n) sites, on thè basis of thè two components separately. Through such a dual relation of similarity it is possible to obtain two graphic arrangements of thè sites.

The sites with thè same environmental components are placed in thè same sector of thè ordination diagram. If thè choice of thè environmental parameters was correct, thè same sites will be placed near in thè sector even in thè ordination diagram on thè basis of thè animals communities.

All thè possible variations between thè positions of thè sites in thè two ordination diagrams reflect thè informations to fix thè relative importance of thè patches in thè studied landscape.

83

Fulgione D. et ah - Procedure to study animai communities

The procedure offers reliable results when it is applied to sites with similar environmental typologies, reducing, for those, thè variabilities examined.

Methods

Operative Procedure

We distinguish four steps:

1 . individuation and quantifìcation of thè parameters that describe thè animai communities and thè environmental mosaicism.

2. Valuation of thè similarity relation between all thè animai commu¬ nities of thè (n) sites considered.

3. Valuation of thè similarity relation between thè environmental parameters used by us to describe thè (n) sites.

4. Reading and comparison of two ordination results.

To exemplify thè application of thè procedure, this one was applied at 12 sites of thè Campania Region (South Italy). The birds (during thè breeding season) were been thè animai taxon monitored. This temporary community is quite link with thè environmental (Farina & Meschini, 1985; Morrison, 1986; Colin et al., 1992).

Since there is a great diversity between thè sites in thè structure of considered landscape, it was been sufficient an animals communities description based on thè presence/absence of thè species.

The environmental typologies, in this exemple, founded on vegetatio- nal and man-made impact criteria, was been described likely for thè campanian sites in Milone (1991).

Step one

Birds were counted during spring-summer season (breeding season) through thè IPA modified method (Blondel, 1970).

From list species recorded in each site it was built thè matrix 0/1 (table I). Environmental data, about thè vegetation and thè presence/ab¬ sence of human elements, were collected, during field survey, recording their percentage within a 200 mts range. In thè table II it is showed thè environmental matrix.

84

Boll. Soc. Natur. Napoli - Voi 102 (1994-1995)

Step two

To evaluate thè similarity relation between thè 12 communities was been utilized thè multivariate analysis method (Principal Component Ana- lisys) using thè statistic package SYNTAX IV for IBM computer (Podani 1991). This method organizes thè sites ordination, based on species infor- mations, in a space of ordination reduced to three dimensions (Digby & Kempton, 1985). In thè figure 1 it is showed thè ordination of thè twelve animai communities censused.

Table I - Heterogeneity of thè 12 sites expressed by thè percentage of thè typologies presents around (radius: 200 mts).

envir. typologies/sites	1	2	3	4	5	6	7	8	9	10	11	12
sea reef	25
tree-lined garden	50	15	70					15
compact building		80	10
orchard		5	20					25		90
area with farms						5		20				10
growing with cereals				50	50	55	40	30			20	40
area with bushers	25			40	10				10		30
scattered wood young wood overgrown with bushes				10	40	30	10				50	10
riparian shrubland						10	40		20
fresh water courses peppled							10	10	30
area with man-made impact areas with olive									40	10		40

Step three

To evaluate thè similarity relation between thè twelve environmental typologies was been utilized thè Agglomerative Analysis. The cluster (fig. 2) was constructed from similarity matrix obtained with UPGMA method and by Sorensen index.

Step four

Very delicate step is thè comparison between thè ordination diagram and thè cluster. The spatial distribution of birds depends by many factors

Fulgione D. et al. - Procedure to study animai communitìes			85
Table II - Zoocenotic informations expressed by presence/absence of birds species
species/sites	1	2	3	4	5	6	7	8	9	10	11	12
Milvus milvus	0	0	0	0	0	1	0	0	0	0	0	0
Buteo buteo	0	0	0	0	0	1	0	0	0	0	0	0
Accipiter nisus	0	0	0	0	0	0	0	0	0	0	1	0
Charadrius dubius	0	0	0	0	0	0	0	0	1	0	0	0
Columba palumbus	0	0	0	0	0	0	0	0	0	0	1	0
Streptopelia turtur	0	0	0	0	0	0	1	0	0	0	0	0
Attiene noctua	0	0	0	0	0	0	0	1	0	0	0	0
Upupa epops	0	0	0	0	0	0	0	0	0	0	1	0
Picus viridis	0	0	0	0	1	1	0	0	0	0	1	0
Lullula arborea	0	0	0	1	0	0	0	0	0	0	0	0
Motacilla alba	1	0	0	0	0	1	1	0	0	0	0	0
Saxicola torquata	0	0	0	0	0	0	0	0	0	0	0	1
Luscinia megarhynchos	0	0	0	0	0	0	0	1	1	1	0	0
Turdus merula	0	1	1	1	0	0	1	1	1	0	1	0
Cettia cetti	0	0	0	0	0	0	0	1	0	0	0	1
Cisticola juncidis	0	0	0	0	0	1	0	1	0	0	0	1
Sylvia atricapilla	0	1	1	1	1	1	1	1	1	1	1	0
Phylloscopus collybita	0	0	0	0	0	0	0	0	0	0	1	0
Troglodytes troglodytes	0	0	0	0	0	0	1	0	0	0	0	0
Sitta europaea	0	0	0	0	1	0	0	0	0	0	0	0
Muscicapa striata	1	0	0	0	0	0	0	0	0	0	0	0
Ficedula sp.	0	0	0	0	0	0	0	0	0	0	1	0
Lanius collurio	0	0	0	1	0	0	1	0	0	0	0	1
Lanius senator	0	0	0	0	0	0	0	0	0	0	1	0
Par us major	0	0	1	0	1	0	1	0	0	0	0	0
Parus caeruleus	0	1	0	0	1	0	0	0	0	0	0	0
Aegithalos caudatus	0	0	0	0	0	1	1	0	0	0	0	0
Pica pica	0	0	0	0	0	0	1	0	0	0	1	0
Garrulus glandarius	0	0	0	1	0	0	0	0	0	0	1	0
Oriolus oriolus	0	0	0	0	1	0	1	0	0	0	1	0
Corvus c. cornix	0	0	0	0	0	0	0	1	1	0	0	0
Fringilla coelebs	1	0	1	0	1	1	1	0	0	1	1	0
Carduelis carduelis	1	0	0	0	0	0	0	1	1	0	0	0
Carduelis chloris	0	0	0	0	0	0	0	1	0	1	0	1
Serinus serinus	0	0	0	0	0	0	0	1	1	1	0	1
Passer montanus	0	0	0	0	0	0	0	1	0	1	0	1
Passer italiae	1	1	1	0	1	1	0	1	1	1	0	1
Milaria calandra	0	0	0	1	0	0	0	0	0	0	0	0
Emberiza cirlus	0	0	0	1	1	1	1	0	0	1	1	0

86

Boll Soc. Natta . Napoli - Voi. 102 (1 994-1 995)

PC2

SBGR * SUB-3 +SUB-2 x SUB- 1

Figure 1 - Three dimensionai plot of thè 12 sites based on thè PCA analysis at 39 variables (species). The symbols represent sub grò up one (X), sub group two (*), and soub-group three (+), indicated by cluster analysis.

in accordance with very complex patterns. For exarnple. thè number and thè position of thè individuals in thè habitat largely depend from thè vegetation stmcture (MacArthur et al., 1962; Blondel et al ., 1973).

This step is very important to vaine which factors are really involved in thè organization of animals communities.

In this step it is possible to formulate hypothesis about thè importance of e very patches considering which level of precision was described thè environmental rnosaic ,

Results

In fìg 2 thè diagram shows three sub-group with 35.0% similarity score: sites 6,7, 9, and 12 (I); sites 2, 3, 8 and 10 (II); sites 1, 4, 5 and 11

(III).

Fulgione D. et al - Procedure to study animai communities

I) The sites 6, 7, 9 and 12 are characterized by thè presences of wetlands ( riparian shrubland, fresh water courses pebbled ); sites 6, 7, and 12 have patches with open corn field with scattered houses ; a man-made impact involves thè sites 9 and 12.

10%

°5%

50%

70%

k

A

Figure 2 - Dendrogram constructed using thè similarity between thè environmen- tal typologies of thè 12 sites. The similarity distance is valued using thè Sòrensen Index and thè average linkage between thè groups (UPGMA method).

II) A common characteristic between sites 2, 3, 8 and 10 is thè orchard but each having different extension. The tree-lined garden are present in 2, 3 and 8 sites; thè compact building is present only in 2 and 3.

Ili) The similarity between 1, 4, 5 and 1 1 is probably due to thè presence of patches overgrown with bushes ; thè open corn field with scattered houses is present in thè sites 4, 5 and 11; in this four sites thè uncommon patches are various, thè site 1 is thè more variable.

88

Boll Soc . Matur. Napoli - Tol. 102 (i£94~Ì995)

The diagram of figure 1 (communities ordination) does not show cìearly this three sub-group. The three principaì component carri es thè 53,3% of thè total variance expressed by species.

The sites 6, 7, 9 and 12 are separated in two pairs, 6=7 and 9-12.

The sub-group 2, 3, 8 and 10 reflects, with high est faithfulness, thè similarity relationship dictated by thè environmental typologies (see thè cluster in fìg. 2).

The sites 1,4,5 and 1 1 delimit a precise sector of thè plain but express a great variability.

Discussiom

The sub groups shown by thè cluster analysis are not always found fully in animals communities ordinations (see para) lei ely figg. 1 arid 2).

This aspect is due to thè scarce ecological importance of thè common patches, inside thè sub-group, which thè cluster it was structuring.

In thè first sub-group, thè presence of wet habitats in thè sites 6, 7, 9 characterizes only thè birds communities of thè first two sites. In thè site 9 thè man-made impact has a very strong effe et on thè bird community. The position of thè site 9 is near to that one of thè site 12 in thè ordination diagram of animals communities: this is probably due to thè presence of relevant percentage of man-made patches (40%). The site 12 has also patches common with those ones of thè sites 6 and 7 (principally open corn fìeld with scattered houses ) but these patches have less importance than those ones with man-made impact present in this site too.

In thè second sub-group, thè mosaic composition and thè birds com¬ munities of thè sites 2 and 3 are similar. It may be that their common patche, compact building , it's more important than others for thè birds. Indeed thè patches with orchard, present in all sub-group sites (2, 3, 8 and 10), do not seem to contribute to create thè similarity between communi¬ ties. The percentage presence of thè orchard is great in thè site 10 (90%) but it is not different between thè site 8 and thè site 3 (25% and 20% respectively). Other environmental elements, as thè farms , could render thè communities of thè site 8 similar to this one of thè site 2 or that one of thè site 3. But, this fact does not occur because, probably, this element “weighs” less of thè viver present in thè site 8 only.

Note that thè presence of patches with fresh water in thè sites 6, 7, 8, 9 is associated with patches with different typologies (more naturale in thè

89

Fulgione D. et al - Procedure to study animai communìtìes

first two than in thè second ones), so that thè animai communities are very different.

The third sub-group 1, 4, 5 and 1 1 is an interessant case that shows thè relative importance of thè wood.

In thè sites 4 and 5 it is present a young wood or a scattered wood :

their communities is largely determined by thè presence of thè typologies “ over grown with bushes” and “open corn field with scattered houses” (note thè position of thè sites 4 and 5 in thè ordination diagram).

On thè contrary thè site 11 with mature wood imposes a typical communities with Accipiter nisus, Columha palumbus, Upupa epops, Fice- dula sp.

In conclusion thè multivariate procedure used allows to characterize thè environmental gradient (Massa and Lo Valvo, 1994).

For example, all thè sites with patches characterized by man-made impact was distributed along thè CP2 axis. The sites 6 and 7 can be cosidered thè more naturai areas. In thè site 1 thè anthropic presence is light; in thè site 3, thè 20% is covered by orchard and thè 10% by building ; thè site 2 is characterized by thè 80% of building. The site 10 is structured on two patches, where thè largest is an intensive monoculture of kiwi, an exotic plant, more frequented by man and with a great pressure of pesti¬ cide. This fact shows thè importance of thè landscape heterogeneity on thè biological diversity (Green, 1989; Naveh, 1991; Hansen & di Castri, 1992). Finally, we have thè position with less scores for thè sites 9 and 12, linked to areas with industries or mines.

Therefore, thè analytical procedure suggested can be applied with different levels of accuracy. Every level is imposed by operator beginning from field data census, using selected indicators as thè communities. It is important to note that thè correlation between thè environmental factors and thè animai communities (step four) is strongly related with environ¬ mental and zoocenotic data.

Acknowledgements

This work have received grants from M.U.R.S.T. 60%.

90 : Boll: Soc, Matur . Napoli • Voi 02 ' (1 994-1 995 %

REFERENCES

Blondel, J., C. Ferry, B. Frochot, 1970. La méthode des indices ponctuels d'abon- dance (I.P.A.) ou des relevés d'avifaune par station d'écoute, Alauda, 38: 55-71.

Blondel, J , C. Ferry, B. Frochot, 1973. Avifaune et végétations essai d'analyse de la diversité, Alauda , 41: 63-84.

Colin, J.B., D.B. Neil, D.A. Hill, 1992. Bird Census Technique , British Trust for Ornithology and Royal Society for Protection of Birds, Academic Press Ine. San Diego, CA.

Digby, P.G.N., R.A. Kempton, 1985. Multiv ariate analysis of ecological communi - ties, Populations and Community Biology Series, Principal Editor, New York, U.S.A.

Farina, A., E. Meschini, 1985. Le comunità di uccelli come indicatori ecologici, Fasola, M. (red.). Atti III Conv. ital Orn. pp: 185-190.

Green, B.H., 1989. Agricultural impaets on thè rural environment. J. Appi Eco - logy, 26: 793-802.

Hansen, A.J., F, di Castri, 1992. Landscape Boundaries, Consequences for Biotic Dìversity and Ecological Flow , Springer-Verlag New York, U.S.A.

MacAkthur, R.H., J.W. MacArthur, I. Preer, 1962. On birds species diversity. IL Prediction of bird census from habitat measurament. American Naturatisi., 96: 167-174.

Massa, B., M. Lo Valvo, 1994. Breedeng bird communities along insular mediter- ranean gradients. Anim. Biol, 3: 15-29.

Melone, M., 1991. Alcuni aspetti metodologici nell'analisi delle comunità orniti che dell'Alta Valle del Seie. Atti del Seminario di Studio “ L’ecologia della eterogeneità “ a cura di A. Farina, Ed. Zara, pp: 31-35.

Morrison, M.L., 1986. Bird population as indicators of environmental change, Current Ornithology , 3: 429-451.

Naveh, Z., 1991. Biodiversità ed eterogeneità ecologica dei rilievi mediterranei. Atti del Seminario di Studio “L’ecologia della eterogeneità " a cura di A. Farina, Ed. Zara, pp: 47-60.

Podani, J. 1991. SYNTAX IV, Computer program for data analysis in ecology and systematic. In: Feoli E. e Orloci L. (eds.), “Computer Assisted Vegetation Analysis”, pp: 437-452. Kluwer Acad. pubi., NL.

Terborgh, L, 1970. Distribution on thè environmental gradients: theory and p re- li mmary interpretation of thè distributional patterns in thè avifauna of thè Cordillera Vilcabamba, Perù. Ecology , 52 1:23-40.

Boll Soc , Natur . Napoli - Voi 102 (1994-1995): 91-98 91

Fire in thè Mediterranean Scrub: Evaluation of thè Impact in thè Licola-Castelvolturno Reserve (Ce) Through

thè Birds

Fulgione D., Rusch C.E. and Milone M.

Dipartimento di Zoologia Università di Napoli Federico II Via Mezzocannone 8, 80134 Napoli

Key words: Fire, Birds, Mediterranean Scrub

Abstract. The mediterranean scrub is a quite heterogeneus ecosystem with a vegetation characterized by more vulnerable species together with others defined resistant to fire that represents a very important element of thè disturbance. In this vegetation inhabit many species of passerines. We have studied thè effects of thè fire partially affecting an area of mediterranean scrub, thè Reserve Licola-Castel¬ volturno (268 Ha), with regard to thè dynamism of ornithical communities. The fire concerned thè above 6-7 Ha during July 1990. The ornithical data have been recorded frorn Aprii 1990 to July 1991. The data processing point a limited impact of thè fire on thè area of study. The similarity between thè species captured in June 1990 and those captured during June 1991 suggests that thè breeding community has been altered very little. After thè fire thè richness and biomass monthly fluctuations are remarkable but thè equitability is rather Constant, even if it is possible to see a little increase. The wintering community seems less stable than thè migration-breeding community even if thè wintering biomass is more abun- dant of thè migrating one. Those variations could be correlated with modifications of fructification periods of mediterranean scrub species.

Riassunto. Il fuoco in ambiente a macchia mediterranea: valutazione dell’impatto nella riserva di Licola-Castelvolturno, utilizzando la classe degli uccelli.

La macchia mediterranea ha un ecosistema molto variegato con specie vulne¬ rabili mescolate ad altre più resistenti all’impatto del fuoco il quale rappresenta il maggiore fattore di disturbo. A questo ambiente sono legate molte specie di uccelli passeriformi. Il nostro lavoro riporta gli effetti che un incendio parziale, della Riserva Naturale di Licola-Castelvolturno (268 Ha), ha provocato sul dinami¬ smo stagionale delle comunità di uccelli. L’incendio ha interessato una superficie di 6-7 Ha durante luglio 1990. I dati sono stati raccolti da aprile 1990 a luglio 1991 e le elaborazioni evidenziano un limitato impatto del fuoco. La similarità tra le

Received 3.3.95, accepted 28.11.95

92 Boll Soc . Afafttr. Nàpoli - Voi 102 '{ J 994-1 995 ), ' ' |

comunità nidificanti di giugno 1990 e quelle di giugno 1991 avvalorano tale ipotesi. Dopo l’incendio la fluttuazione mensile dei valori di ricchezza e di bio¬ massa subiscono un incremento, mentre l’equipartizione evidenzia valori più stabili. Le comunità dei mesi invernali evidenziano una stabilità maggiore di quelle dei periodi di migrazione e di riproduzione. Durante il periodo invernale si notano anche elevati valori di biomassa. Alcune di queste modalità possono essere messe in relazione ai modificati periodi di fruttificazione delle essenze della macchia mediterranea.

Introduction

Fire represents a very important element of thè disturbance in a naturai ecosystem, especially in thè mediterranean region. In thè latest twenty years thè fires bave interested in Italy 2.536.907 Ha (which 1.400.691 Ha in thè scrub), in particular thè Campania Region was cov- ered by 19.903 fires interesting 136.319 Ha, of which 62.419 Ha were in notwooded land (Ministero dell’Ambiente, 1992). The effects of thè fire vary according to a number of parameters, peculiar to thè concerned System. It is important, for istance, thè vegetation type (Turner and Romme, 1994), its spatial structure (Lambertini, 1989), thè distribution of those sites more susceptible to thè fire whìtin thè generai pattern of thè landscape (O’Neill et al., 1992) and thè successional seres of thè interessed community (Swanson et al., 1992).

Recently, thanks to thè coming of Landscape Ecology, thè importance of thè scale in which it was we measured thè effe et of thè fire has been understood. In fact, if thè fire is referred to a very small spatial scale, it may represent a destructive event; however if its effects were evaluated in relation to a very large scale, it may even become a mechanism allowing a more various dynamism of thè different forestal components (Farina, 1993).

What above seems proved also as far as specific richness and abun- dance of birds, reptiles and isopods are concerned; it is also most relevant for thè conservation of biodiversity aimed at thè hìghest value of methasta- bility (Naveh, 1991).

The mediterranean scrub is a quite heterogeneus ecosystem with a vegetation characterized by a greater number of evergreen species. The volatile oils secreted by thè many plants make thè vegetation easily flammable. More vulnerable species are present together with others defin ed resistant. Most of them are even whetted in their growth (Cox et

93

Fulgione D. et al. - Fire in thè Mediterranean Scrub

al., 1989). In this vegetation inhabit many species of passerines; some of them find shelter and nourishment during thè migrations and thè winter- ing period, others select this area as comfortables sites for nesting.

This study is to be included in thè framework of thè activities brought on by thè Department of Zoology of Naples, it has been carried on in collaboration with thè foresters of "Corpo Forestale dello Stato”. It aims at evaluating thè effects of thè fire, partially affecting an area of mediter¬ ranean scrub, with regard to thè dynamism of ornithical communities.

Study area

The concerned area covers thè retrodunal strip, its extension is 268 Ha (from 1 to 8 m above sea-level). It is a preserved area (Licola-Castel- volturno, CE) run by thè "Corpo Forestale dello Stato” (fig. 1).

The reserve, formed by many consolidate dunes near thè sea, is situated in thè Volturno's river plain.

Here it is possible to find various elements of thè mediterranean scrub, particularly Phillyrea latifolia, Pistacia lentiscus, Myrtus communis , Erica arborea , Rhamnus alaternus. The more inner area is founded by an artificial pine wood of Pinus pinea, Pinus pinaster, Pinus halepensis, about 50 years old (de Filippo et al., 1982).

The fire concerned thè above 6-7 Ha along thè strip of scrub charac- terized by ilex and pines during July 1990.

Materials and methods

The ornithical data have been recorded from Aprii 1990 to July 1991, excepted thè month of thè fire and thè following one (July 1990 and August 1990).

Captures with mist - nets placed along a 300 m transect have been carried out. This method has allowed to sample with good approximation thè community of birds breeding in thè area and halting during thè migration and winter periods. The caught birds are weighed with a pesola (0,1 g accuracy). At thè same time a record of thè fructification periods of thè most common plants growing in thè area has been taken, in order to measure thè seasonal variability of nourishment offered by thè site.

The data processing, realized with informatic packages for multivari- ated statistic analysis (SYNTAX IV of Podani, 1991) has been drawn from

94

Boll Soc . Natur. Napoli - Voi 102 ( 1994-1995 )

a data matrix given by thè frequencies relevant to each specie throughout thè months. A matrix of similarities among thè communities, for each month has been drawn up.

The following stage has been thè evaluation of similarities among thè months, following thè Complete Linkage strategy and thè Euclidean Dis- tance estimate.

\

\

Lìcola-Castelvotumo y resene ^

D

Naphss

\

y3

A.

"'Vv

$

<r

l

V

i

k

V

V

\

Figure 1 - The study area: Reserve of Licola-Castelvolturno (268 Ha).

The results have been put into evidence through a cluster.

The index of equitability “J” (Pielou, 1966) has been calculed. Finally, thè annual course has been laid out together with “S”, thè richness of thè species, and biomass (BM).

Fulgione D. et al » Fire in thè Mediterranean Scrub 95

' '

Results

The cluster analysis (fìg. 2) points out a clear-cut distinction between thè spring/summer months and thè autumn/winther ones. The similarity between thè two groups is 52%. The community of February, even though belonging to first group, parts frorn it at a 76% level of similarity. The communities of thè reproductive period point out a level of similarity higher than 90% if considered before (June 1990) and after thè fire (June 1991).

SIMILARITY 0,95 0,90 0,83 0,76

i 1

Aprii 90 July 91 May 90 March 91 May 91 Aprii 91 June 90 June 91 February 91 — Septemb.90 — October 90 -i Novemb.90 -* Decemb.90 — JanuaiyPl —

b— i

Figure 2 - Similarities among thè monthly birds communities. Cluster Analysis carrying out following thè Complete Linkage strategy and thè Euclidean Distance.

Figure 3 shows thè high richness and a relevant biomass during thè wintering and thè spring migration period with a minimum in thè late winter (February), but thè equitability is rather Constant, even if it is

96 Boll. Soc. Natur. Napoli - Voi 102 ( 1994-1995 )

possible to see a little increase after thè fire. After thè fire thè richness and biomass fluctuations could be remarkable.

I — X— S ■ ■ BM A 7]

Figure 3 - Annual modifications of three communities indexes: thè equitability “J” (triangles); thè species richness “S” (stars); thè biomass “BM” (squares).

In Table I are shown thè periods of fructification, obtained through our examination of thè berries. These ones were compared with those ones described in a nearest area by Terracciano (1910).

Table 1 - Periods of fructification in thè most common plants growing in thè area (data compared with those ones described in a nearest area by Terracciano, 1910).

SPECIES

Fructif. periods: TERRACCIANO

OUR OBSERVATIONS

Rhamnus alaternus Pistacia lentiscus Myrtus communis Smilax aspera Phillyrea sp. Asparag. acutifolius

July - August September - October October - November October - November Aprii - June October - November

May - August September - December October - January January - November June - July October - December

Discussion

The data processing point a limited impact of thè fire on thè area of study. The usuai seasonal turn-over of ornithical communities seems re-

97

Fulgione D. et al. - Fire in thè Mediterranean Scrub

spected, in fact, there is a clear-cut distinction between spring/summer months and autumn/winter ones, February being a typical month of tran- sition in relation to thè phenology of thè species considered. The similarity (fìg. 2) between thè species captured in June 1990 and those captured during June 1991 suggests that thè breeding community has been altered very little.

The observing annual fluctuations of index (fig. 3) could be related to thè limited extention of thè fire because this one has caused only a small patch in thè whole ecosystem. This patch has been characterized by an earlier successional sere as to thè surrounding environment.

The peculiar process of mosaicism due to thè locai effe et of thè fire (Turner and Romme, 1994) creates areas of regressed habitats, dephased with respect to thè successional sere of thè whole ecosistem increasing thè richness appreciably (Grubb, 1977; Connell, 1978; Huston, 1979). This is probably thè spring 1991 case (see fig. 3), here thè ratio between thè successional sere in thè patch and that one in thè whole ecosystem is more remarkable. It would then seem logicai to observe thè impact on a greater scale in order to study its reai effeets. It is important to note thè increase of “J”, from thè wintering period to thè migration-breeding period, after thè fire disturbance. The wintering community seems less stable than thè migration-breeding community even if thè wintering biomass is more abundant of thè migrating one.

A sympton of regression of thè succession in thè patch could be thè generai shift of thè fructification of almost all thè species, compared with thè data of Terracciano (1910). Particularly, it was observed that Rhamnus alaternus, Pistacia lentiscus, Smilax aspera and Myrtus communis comes earlier their fructification (these specie seems to be a more large period of that known), on thè contrary Phillyrea sp. postpones their fructification.

These variations could be related with thè two peaks of richness. The productivity of Pistacia lentiscus and Myrtus communis could be corre- lated with thè bird richness and biomass during thè wintering while thè productivity of Ramnus alaternus could be related with thè migration period.

Now becomes important to known if all these observations are wild- fire effeets. As a consequence, in such perturbation-dependent landscapes a possible management goal should not be thè achievement of a homeo- static steady-state climax, leading to a dense and a monotonous woody cover through non-interference and complete protection, but thè mainte- nance of thè homeorhetic flow equilibrium through Constant, well-con- trolled intervention by “optimum” defoliation pressure (Naveh, 1991).

98

Boll. Soc . Natur. Napoli - Voi. 102 (1994-1995)

Acknowledgements

We would like to thank Dott. Nicola Di Fusco director of Naturai Reserves in Campania of thè Ministry of Agriculture politics ex-ASFD. We are grateful to Dr. G. de Filippo for his help during thè ringing operations. This work was funded by grant from thè Italian Ministero dell'Università e della Ricerca Scientifica e Tecnologica (MURST).

REFERENCES

Connell, J. H., 1978. Diversity in tropical rainforests and coral reef. Science, 199: 1302=1310.

Cox, B., Moore, P.D., Whitfield, P., 1989. The atlas of thè living world. Marshall Editions Limited, London.

de Filippo, G., Fraissinet, M., Grassi, G., Kalby, M., Milone, M., (a cura di), 1982. Proposte per l’istituzione di Parchi e Riserve Naturali in Campania, pp. 77-81. Regione Campania, Napoli.

Farina, A., 1993. L’ecologia dei sistemi ambientali. CLUEP Editore, Padova. Grubb, P.,1977. The maintenance of species richness in plant communities: thè importance of thè regeneration niche. Biologie al Reviews. 52: 107-145. Huston, M., 1979. A generai hypotesis of species diversity. American Naturalist. 113:81-101.

Lambertini, M., Faralli, U., 1989. Ecologia del fuoco in un ambiente a macchia mediterranea (Monte Argentario, Toscana): struttura ed evoluzione delle co¬ munità ornitiche. Atti V Convegno Italiano di Ornitologia, Suppl. Rie. Biol. Selvaggina, 17: 7-10.

Ministero dell'Ambiente, 1992. Relazione sullo stato dell’ambiente. Istituto Poli¬ grafico Zecca dello Stato, Roma.

Naveh, Z., 1991. Biodiversità ed eterogeneità ecologica nei rilievi mediterranei. Atti del seminario di studi “ L’ecologia della eterogeneità” , a cura di A. Farina, Ed. Zara, pp: 47-60.

O’Neill, R.V., Gardner, R.H., Turner, M.G., Romme, W.H., 1992. Epidemiology theory and disturbance spread on landscape. Landscape Ecology 7: 19-26. Pielou, E.C., 1966. The measurament of diversity in different types of biological collection. J. Theoret. Biol., 13: 131-144.

Podani, J., 1991. Syntax IV. Computer Programs for data Analisys in Ecology and Systematics. In E. Feoli and L. Orloci (eds.), Computer Assisted Vegetation Analysis, 437-452. Kluwer Academic Publishers. Netherlands.

Swanson, F.J., Wondzell, S.M., Grant, G.E., 1992. Landforms, disturbance and ecotones. In: Landscape boundaries, Hansen & di Castri (eds.), Springer-Ver- lag, New York.

Terracciano, N., 1910. La flora dei Campi Flegrei. Atti Ist. Incor. Napoli.

Turner, M.G., Romme, W.H., 1994. Landscape dynamics in crown fire ecosystems. Landscape Ecology 9: 59-77 .

CONTRIBUTI IN ITALIANO

Boll Soc . Natur. Napoli - Voi. 102 (1994-1995): 101-116 101

. HHHIIi \ m I Hi H

Il Partial Least Squares e l'Analisi in Componenti Principali rispetto ad un sottospazio di riferimento. Un'applicazione per la rilevazione di inquinamento di ambienti fluviali

S. Scippacercola, P. Amenta e L. D’Ambra

Department of Matematics and Statistics University Federico II of Naples Monte Sant’Angelo, Via Cinthia, 80126 Naples - Italy

Key words: multicollinearity, partial least squares, Principal components, pollution

Riassunto. Negli ultimi anni la letteratura specializzata ha mostrato un inte¬ resse notevole a metodologie in cui è presente la collinearità. A tal fine, per scopi predittivi, è stato proposto, negli anni più recenti, un metodo con il nome di PLS (Partial Least Squares) (Wold, 1966) che si è rivelato molto utile in chemiometria.

Nel presente lavoro, dopo un confronto con altri metodi, viene suggerito l’uso del PLS come metodologia alternativa a tecniche fattoriali vincolate per analizzare congiuntamente le variabili anche in presenza di multicollinearità.

Un'applicazione sull’inquinamento di ambienti fluviali in Italia, consente di verificare la validità del metodo e di fornire, sotto certe condizioni, una interpre¬ tazione più puntuale rispetto ad altri approcci.

Summary. The Partial Least Squares and thè Principal Components Analysis on a reference subspace. The italian rivers pollution.

The Partial Least Squares (PLS) (Wold, 1966) is a predictive method very useful in case of multicollinearity, mainly in chemiometry. In this work we de- scribe and compare PLS with other methods proposed in literature.

Let X (n, p) be a data matrix with p explicative variables observed on n individuai. Let y (n, 1) be a response variable. The classic regression model y = X{3 + u (with J Q (p, 1) and u (n, 1) stochastic variables) estimates thè regression parameters by means of thè least squares method (fi = (X'X) X'y).

If thè determinant of X'X is dose to zero we have thè multicollinearity. In this case thè inverse matrix of X'X does not exist and thè regression coefficients are strongly unstables. Furthermore we could note an incoherence in thè regressor signs. In thè regression, thè main problem arises in thè estimation of jS if p > n.

Other approaches are thè Ridge Regression (Hoerl-Kennard, 1970), thè Latent Root Regression Analysis (Webster-Gunst-Mason, 1974), thè Principal Compo¬ nents Regression (Massey, 1965) and thè PLS.

Received 23.10.94, accepted 6.12.95

102

Boll Soc . Natur. Napoli - Voi 102 ( 19944995 )

The PLS proposed by Wold (1966, 1975, 1989) exists in two different versions: thè PLS1, if we consider one response variable, and thè PLS2 if we have more than one response variables.

In PLS2 we perform a singular values decomposition of thè variance - cova- riance matrix X'Y where Y is of order (n, q). The System of weights for thè matrices Y and X are thè left and right eigenvectors associated to thè highest modulus eigenvalue, respectively.

The PLS2 criterion is to research thè th (linear combination of X) and qh (linear combination of Y) components in order to maximize thè covariance between th and qh. If r represents thè correlation coefficient between th and qh, we have that Cov(th, q,J = r yjVar(th) Var(qh).

In order to obtain thè maximum of this term we have to maximize Var(th) and Var(qh) simultaneously. This “coinertia criterion” gives us coherent results com- pared with thè results of thè alternative methodologies. Other methods leave out thè variables judged “redundant”. The PLS method takes into account all thè variables providing an higher degree of information.

A different approach for studying thè asymmetrical relationship between two different sets of variables X and Y is thè Principal Components Analysis on a reference subspace (ACPR) (D’Ambra and Lauro, 1982). The aim of this method is to analyze thè structure of thè former set of variables with respect to thè latter. We see as ACPR maximizes thè projected inertia of response variables with respect to explicative variables while PLS maximizes thè coinertia between thè two sets. A multicollinearity problem can arise in ACPR. PLS can be considered as a useful tool to avoid this problem.

In this paper thè PLS is applied on a reai case (thè italian rivers pollution) to verify thè validity, thè characteristic and thè advantages of thè method in relation with ACPR in presence of multicollinearity.

The easy geometrie interpretation and thè missing data treatment characteri- zes thè PLS methods as a powerfull tools for descriptive and predictive purposes.

Introduzione

Considerata una matrice di dati X (n, p) contenente p variabili espli¬ cative osservate su n individui ed una variabile di risposta y (n, 1), il modello classico di regressione y = Xf3+u (con fi(n, 1) ed u(n, 1) variabile casuale stocastica) porta a stimare i coefficienti di regressione j3 mediante il metodo dei minimi quadrati j8 = (X'X)-1 X'y.

Se il determinante della matrice X'X è pari a 0 vi è multicollinearità perfetta o, caso più frequente, se il determinante di X'X — 0 si è in presenza di multicollinearità. In questi casi la matrice (X'X)1 non esiste ed i coefficienti di regressione mostrano una elevata instabilità e, cosa ancora più grave, spesso si possono evidenziare, a posteriori, incoerenze nel segno dei regressori.

Scippaèefcòla & et al, - Il Partial 'beasi Sguares.

Nell'ambito della regressione il principale problema è la stima di 0 se p > n. Le proposte più comuni sono la ridge regression (Hoerl-Kennard, 1970), l'analisi delle radici latenti (Webster-Gunst-Mason, 1974), la regres¬ sione sulle componenti principali (Massey, 1965), ed, a questa lista, si può aggiungere il PLS. Altre proposte metodologiche, in caso di multicollinea- rità, portano ad escludere dal contesto interpretativo alcune variabili che concorrono alla spiegazione del fenomeno. L’esclusione di variabili forte¬ mente correlate pregiudica l'analisi globale del fenomeno e, come sotto- linea Martens (1987), la presenza di multicollinearità dovrebbe essere un vantaggio e non un problema da risolvere.

Il PLS, per il particolare criterio di co-inerzia utilizzato, riesce a fornire l'interpretazione di molti casi che, invece, con altre metodologie risultano non aderenti alla realtà sperimentale. Il PLS nasce come algo¬ ritmo e, benché ripetutamente citato in letteratura, nonostante i suoi evidenti vantaggi come metodologia alternativa nei casi di multìcollinea- rità, ha trovato scarse applicazioni.

Scopo del presente lavoro è di descrivere il metodo PLS (par. 2), di confrontarlo con altre metodologie e di analizzare i risultati di una appli¬ cazione del PLS con l’Analisi in componenti principali rispetto ad un sottospazio di riferimento (ACPR) (D’Ambra- Lauro, 1982) su alcuni para¬ metri caratterizzanti i fiumi d’Italia per rilevarne il grado di inquinamento (par. 3).

Metodi

- Il Metodo PLS

Date p variabili esplicative, il PLS proposto da Wold (1966, 1975, 1989) esiste in due distinte versioni: PLS1 e PLS2, quando si considerano, rispettivamente, lek variabili di risposta. Vengono, di seguito, breve¬ mente illustrati i metodi PLS1 e PLS2:

- Il Metodo PLS1

1 . Considerata una matrice di dati X (n,p) contenente le determinazioni di p variabili su n osservazioni, detta y (n,. ì) la variabile di risposta, ed indicato, rispettivamente, con X ed y la. matrice e lo scalare delle medie, si calcolano gli scarti dalle medie (resìdui): X0 = X - X e y0 = y - l y, dove l = [ 1 ... 1 ]. Per i = 1,2...... si eseguono i passi da (2) a

(5).

104

Boll. Soc. Natur. Napoli - Voi. 102 (1994-1995)

2. Si calcolano i vettori q (n, 1) con una opportuna combinazione li¬ neare dei residui X(i_1} del passo precedente; t(i) = X(i_1}w\i)f con w(i) — y' (i-i)X(i-i). I coefficienti w(i) sono, quindi, calcolati usando la cova¬ rianza tra X e y.

3. Si determinano i coefficienti p{ (1 ,p) (per le p variabili) e (per le y) regredendo, con il metodo dei minimi quadrati, le X e le y su tif otte¬ nendo:

= t'JCj.! Cov(q,X j_j)

P‘ tfì Var(ti)

= = Covfay,,,;

q' Ui Variti

4. Si calcolano i nuovi residui

Xi = Xì_j - tpi yi = yi-i - tfti

(1)

(2)

5. Una possibile condizione di convergenza è la verifica della stabilità della Var(ti) =

Il metodo fornisce ad ogni passo una rappresentazione bilineare di X ed y, al generico passo k:

X - X + tipn + . . . + tkPk + Xk y = y + 1 1q1 + tkqk + yk

(3)

Dalla (3) si deduce che i residui X{ sono incorrelati con th t2, tk; inoltre si dimostra che i pesi wh w2, wk risultano ortogonali (Helland, 1988). I vettori q sono le proiezioni di X f sulla base generata dalle colonne della matrice WA formata dai coefficienti wq inoltre i regressori p{ sono le proiezioni di X{ sulla base generata dalle righe della matrice dei coeffi¬ cienti q . Ciò consente una interessante rappresentazione geometrica.

Il metodo fornisce, quindi, un insieme di coefficienti w,- ortogonali ed un insieme di vettori p{ di coefficienti non ortogonali che si approssimano con i vettori q ortogonali. L'alternanza di proiezioni su righe e su colonne ha un effetto incrociato a cui può essere conferito il nome di regressione criss-cross (Amenta-Scippacercola, 1994).

105

Scippacercola S. et al • Il Partial Least Squares

Un ulteriore vantaggio è che i parametri possono essere stimati per ogni fattore separatamente poiché sia i vettori w;- sia i risultano essere ortogonali; inoltre, il metodo non richiede nessuna inversione.

- Il Metodo PLS2

Il PLS2 considera q variabili di risposta e procede alla decomposi¬ zione in valori singolari della matrice di varianza e covarianza Cov (X, Y) = X'YY'X attribuendo i pesi alle matrici Y (n, q) ed X (n, p), rispettiva¬ mente, con P autovettore di sinistra e di destra corrispondente all’autova- lore di massimo modulo. Vengono, quindi, calcolati i residui e si procede ad iterare il procedimento con tali residui.

Il criterio del PLS è quello di ricercare le componenti th (combina¬ zione lineare di X) e qh (combinazione lineare di Y) tale che la covarianza tra esse sia massima. Ora, poiché

Cov(th> qh) = r y/Varfth) Var(qh) (4)

dove r rappresenta il coefficiente di correlazione tra 4 e qh> si evince che per massimizzare tale covarianza si devono massimizzare simultanea¬ mente: Var(th), Var(qh) e r(th,qh).

La principale difficoltà del PLS2, riscontrabile, peraltro, in altri me¬ todi è il caso in cui si presentano autovalori uguali, condizione che può generare ambiguità nella scelta. Altra difficoltà è la condizione di conver¬ genza. Esistono, in letteratura, varie proposte in merito come la cross-va- lidation (Stone, 1974; Wold, 1988) ed altri metodi (Martens, Naes, 1987; Hanne, 1987; Hòskuldson, 1988).

In appendice sono riportati gli algoritmi PLS proposti da Wold ed Helland.

ACPR ed il PLS

Indicato con X(n>p) (criterio) ed X(n>q) (esplicative), due matrici asso¬ ciate, rispettivamente, a due insiemi di variabili quantitative osservate su n individui, obiettivo delLACPR è quello di analizzare la struttura della varianza spiegata del primo set di dati rispetto al secondo set di dati. L’approccio consiste nell 'effettuare una Analisi in Componenti Principali delle immagini delle variabili criterio rispetto al sottospazio generato dalle variabili esplicative.

106

Boll Soc . Natur. Napoli - Voi. 102 ( 1994-1995 )

Siano 9ty e 9^x i due sottospazi di 9L1 generati dai vettori linearmente indipendenti x; eXey^ gY (j = 1, ..., q) (k = 1, p) e siano, con ln matrice identità di ordine n,

P,i - \ln - (1 /n)unun] e Po = P^X(X'P^X)-' X'P^

i proiettori ortogonali, rispettivamente, al vettore un = [ 1 1 ... 1 ] ed al sotto¬ spazio 9tx.

Al fine di visualizzare la struttura di dipendenza fra Y e X, l’ACPR ricerca gli assi principali vf (i = 1, ..., h < min (p,q)) di 9tx tale che

max Vf-Y'P^PJP^Yvi soggetto ai vincoli v-,vf = 1 e v-,vf = 0 per i ^ i

V, fi

L'ACPR ricerca un vettore di pesi b che massimizza la quantità

\t't/b'b]

dove t = PoYh e P0 è l'operatore di proiezione ortogonale sul sottospazio di riferimento generato dalle colonne di X. Si può apprezzare la differenza dei criteri tra l'ACPR ed il PLS: l’ACPR massimizza l’inerzia spiegata (proiettata) mentre il PLS massimizza la coinerzia Cov(b'Y'X) in quanto ricerca un vettore di pesi che massimizza la quantità [t'y/b'b] con t = Xb.

Altri metodi ed il PLS

- Nel modello lineare di regressione , utilizzando il metodo dei minimi quadrati, si stimano i coefficienti /? con /3 = (X'X)~l X'y. Per contro la metodologia PLS conduce alla stima dei coefficienti j QPLS con

Ppls = W.CW.X'XW,)-1 WAX'y dove la relazione di ricorrenza è stabilita da

wA+1 = xy - iwdw.nw,)-1 Kxy

con WA matrice dei coefficienti w,.

Scippacercola S. et al, - Il Parimi Least Squares

107

La soluzione individuata con il PLS risulta analoga alla soluzione separata di una Analisi in Componenti Principali (ACP) su X e su Y ma è effettuata in una singola iterazione.

- L' Analisi in Componenti Principali ricerca un vettore b di pesi che massimizza il rapporto

[ft/b'b]

con t = Yb massimizzando, quindi, la varianza Var (Yb); il PLS, invece, ricerca un vettore di pesi che massimizza la quantità

[t'y/b'h]

con t — Xb e dove il numeratore risulta pari alla covarianza ( b'X'y ). Si evince come i criteri dei due metodi portino il primo ad esaminare la struttura mentre il secondo la co-struttura.

- Indichiamo con c* una base unitaria di un sottospazio K-dimensio- nale di uno spazio euclideo p-dimensionale (K < p) e con a=S£=1 akck un vettore di questo sottospazio. Nel caso dei minimi quadrati ordinari (MQO), il sottospazio è definito dal vettore unitario che massimizza la correlazione, al quadrato, fra la variabile di risposta e la corrispondente combinazione lineare delle variabili predittrici:

cMqo = max corr2(v, Xc)

c'c= 1

Nel caso della Ridge Regression (RR) (App. 3), il sottospazio viene generato dal vettore unitario definito dal seguente criterio con X parame¬ tro “ridge”

c rr = max corr2(j, Xc)

c'c = 1

var (y, Xc) var(y, Xc) + X

Il criterio della Principal Component Regression (App. 2) porta a definire una sequenza di sottospazi K-dimensionali ognuno generato dai primi K autovettori associato alla decomposizione in valori singolari della matrice di varianza e covarianza delle variabili predittrici. Ogni ck (1 < k < R) è basato sul criterio, per / = 1, ..., k - 1

c PCR = max var(Xc)

cVc=0 c'c= 1

dove R è il numero di autovalori non nulli.

1

08 Boll. Soc. Natur.

Napoli ■

■ Voi. 102 (1994-1995)

Anche la metodologia PLS produce una sequenza di sottospazi K-di- mensionali generati da vettori unitari e le K soluzioni sono ottenute a

Fosforo

N

N-Nh4

Gap. Corri. 2

Garbon.

Creso. alghe20 Creso. alghelO

Gap. Corri. 1

Gpnduc.

Durezze

Creso. algheSO

Ks

Cu

Ph

Asse 1

Figura 1 - ACPR. Rappresentazione delle variabili

partire da un'adattamento a minimi quadrati della variabile di risposta sulle K corrispondenti combinazioni lineari. Il PLS, anche se utilizza una strategia assai prossima al PCR, se ne differenzia in quanto al criterio di definizione dei sottospazi K-dimensionali e delle combinazioni lineari:

c pLS = max corr2(y, Xc)var(Xc)

c'Vc=0

c'c=l

109

Scippacercola S. et al. - Il Partial Least Squares

Risultati di un'applicazione sull'inquinamento di fiumi

Per confrontare i risultati della metodologia PLS rispetto all'ACPR, sono stati presi in esame 28 determinazioni effettuate su campioni d'acqua prelevati in alcuni fiumi dell’Italia settentrionale (Camusso et al., 1992).

esc.alghelO

Cresc.alghe20

-8

Creso. algheSO

Cap.Com.1

Ph

Conduc.

CarbWVezZa

Gap. Corri. 2

N

Fosforo

N-Nh4

Cu

\

-6

I

-4

Asse 1

Figura 2 - PLS. Rappresentazione delle variabili

Le variabili esplicative considerate sono:

• Conducibilità ( \xS/cm ), pH, Durezza ( oF ), Fosforo totale (ptgP/l), Azoto ammoniacale ( mgN/l ), Azoto totale ( mgN/l ), Carbonio organico di¬ sciolto ( mgC/l )

110

Boll Soc. Natur. Napoli - Voi 102 (1994-1995)

• Due Capacità complessanti ( pimolCu/l ) derivanti da distinti metodi

• Costante di stabilità condizionale (KS) ( l/panol )

• Rame

CM CD

CO CM _

Bardelle

Oglio

Serio Tefdoppio

Po3

oraBaltea

Arda

Tanaro2

Brembo

Tanarol

Agogna

Addai

Dezzo

Scrivia

Sesia

Po2

Chiese

Nure

Tar 3

Anza S. Bernard

Mera

Vsv°2

Toce Ticinol

Adda

Mailer©

-4-2 0 2

Asse 1

Figura 3 - ACPR. Rappresentazione dei fiumi

lino

Come variabili di risposta sono state considerate le crescite di Selena- strum capricornutum dopo 96 ore dall'aggiunta di quantità diverse di Rame (10, 20 o 50 ptg/l) usando, come controllo, altri campioni senza aggiunta di Rame. L'obiettivo della ricerca (Camusso et al., 1992) era di dimostrare che, per valutare l'inquinamento fluviale, è più importante la capacità complessante rispetto alla durezza.

Scippacercola S . et al. - Il Partial Least Squares

111

Sono state eseguite due distinte analisi utilizzando le metodologie ACPR (Fig. 1 e 3) e PLS (Fig. 2 e 4). In Fig. 1 e 2 sono riportati, nel piano fattoriale PII, la rappresentazione dei parametri, rispettivamente, eseguite con 1 'ACPR e con il PLS. Le percentuali di inerzia (co-inerzia) spiegate dal primo asse sono quasi del 70% e del 20% dal secondo asse.

De

Terdc

Poi

Chiese

Tanaro2

Bardello

Arda

Serio

Oglio

Tanarol

Brembo

Po3

Mure

Scrivia

Seri

-6

-4

Mailer©

S. Bernardi

raBaltea

Mera

Toce

ppio

Agogna

ez2^dd^'

Pol?icino2

^aro

Addai

Anza

Ticinol

Asse 1

Figura 4 - PLS. Rappresentazione dei fiumi

I risultati ottenuti con il PLS sono molto più aderenti ai risultati sperimentali. Infatti basta esaminare le diverse posizioni della crescita delle alghe dopo la somministrazione di 50 pig/l di Rame che nell'ACPR non risulta differenziata.

112

Boll Soc . Natur. Napoli - Voi 102 (1994-1995)

Anche per quanto riguarda i fiumi l’ACPR si mostra non adeguata alla situazione sperimentale: basti citare la differente posizione del Nure e del Taro (Fig. 3 e 4) nelle due distinte analisi. L’ACPR tende a collocare in posizione fattoriali anomale i suddetti fiumi.

Inoltre, da una valutazione globale dei risultati e tenuto conto della forte collinearità presente nei dati e dei differenti criteri che soggiaciono alle due metodologie, il PLS conferma ampiamente i risultati sperimen tali, raggiunti dalla precedente analisi (Camusso et al., 1992), rispetto all’ACPR.

Conclusioni

La metodologia PLS nasce per scopi prevalentemente previsionali in presenza di un elevato numero di variabili esplicative fortemente corre¬ late, ovvero quando la matrice (A'A)1 non esiste o nei casi di quasi singolarità od anche quando si rileva a posteriori una incoerenza nei segni dei regressori.

La coerenza dei risultati ottenuti con il PLS appare evidente se si confrontano i risultati di metodologie alternative, e questo è imputabile al criterio di co-inerzia utilizzato (1) (2) (4) basato sulla analisi della co-strut¬ tura di due matrici di dati.

Infine, la facile interpretazione geometrica e la possibilità di trattare missing data conferisce al metodo un ulteriore interessante alternativa per scopi predettivi.

Nota

L’introduzione, le conclusioni e l’applicazione sono stati redatti dai tre autori;

I paragrafi PLS1, ALTRI METODI E PLS e l’APPENDICE 2 sono stati redatti da P. Amenta;

I paragrafi PLS2, ACPR E PLS e le APPENDICI 1 e 3 sono stati redatti da S. Scippacercola.

Scìppacercola S. et al. - Il Partìal Least Squares

113

Bibliografia

Amenta P., S., Scìppacercola, 1994. Regressione Criss-Cross, III. a Riunione scienti¬ fica. Napoli: Dip. Matematica e Statistica, Università studi “Federico II”.

C amusso M., Gaggino, G.F., Marchetti, R., Provini, A., G., Tartari, 1992. Tossicità del rame in acque fluviali a diversa capacità complessante, Acqua- Aria, 1: 31-37.

D'Ambra L., N.C., Lauro, 1982. Analisi in componenti principali in rapporto ad un sottospazio di riferimento. Rivista di Statistica Applicata, 1: 51-67.

Helland, I.S., 1988. On thè structure of partial least squares regression. Comm. Statisi. B Simulation Comput., 17: 581-607.

Helland, I.S., 1990. Partial Least Squares Regression and Statistica! Model. Scand. J. Statisi., 17: 97-114.

Hoerl, A. E., R.W., Kennard, 1970. Ridge Regression: Biased Estimation for No- northogonal Problems. Technometrics, 12: 55-67.

Hoskuldson, A., 1988. PLS Regression methods, J. Chemometrics , 2: 211-220.

Martens, H., 1987. In Helland, I.S., 1990.

Massey, W.F., 1965. Principal Components Regression in Exploratory Statistical Research. Journal of thè American Stat. Ass., 83: 1023-1037.

Stone, M., 1974. Cross-Validatory choice and assessment of statistical predictions

(with discussion). J.R. Statisi. Soc. Series B, 36: 111-133; corrigendum, 1976, 38: 102.

Webster, J.T., Gunst, R.F., R.L., Mason, 1974. Latent Root Regresson Analysis, Technometrics, 16: 513-522.

Wold, H., 1966. Estimation of principal components and related models by itera¬ tive least squares. Multivariate Analysis, P.R. Krishnaiah (ed.), 391-420. New York: Ac. Press.

Wold, H., 1975. Soft modelling by latent variables: Non linear Iterative Partial Least squares approach. In Perspectives in Probahility and Statistics: Papers in honour of Bartelett, J. Cani (ed.), 117-142. London: Ac. Press..

114

Boti, Soc . Natur. Napoli - Voi 102 (19944995)

Appendice 1

Gli algoritmi di Wold e di Helland

Definito con y = {yx ... yq} un set di q variabili di risposta e con x = {xi ... xp} un set di p variabili predatrici misurate su un set di n osservazioni {yh Xi} = {yu ... yqi) xu ... xpi}, per i = 1, ..., n, l’algoritmo proposto da Wold (1966), che non rientra tra le procedure dei modelli lineari, produce una sequenza di modelli {y*} ( k = 0, ..., R), ad ogni passo del ciclo. Il modello {y*} che minimizza il criterio prosposto da Stone (1974) viene scelto come soluzione dell'algoritmo PLS. Ad ogni passo k si effettua una regressione parziale dei residui yk_x rispetto ai residui x*_i ottenuti al passo precedente. Per k = 0 i residui yex vengono inizializzati con i rispettivi dati originali standardizzati. La regressione parziale consiste nel calcolare il vettore contenente le covarianze w* per formare una combinazione lineare zk dei residui x. I residui y vengono quindi regrediti su questa nuova combina¬ zione lineare ed il risultato è aggiunto al modello e successivamente sottratto dai residui y per formare i nuovi residui yk da utilizzare nei passi seguenti. Vengono calcolati allora i nuovi residui xk sottraendo da xk_x la sua proiezione su zk. La condizione di arresto interromperà l’algoritmo dopo R cicli dove R è il rango della matrice dei prodotti scalari V.

Fin dalla sua prima formulazione diverse soluzioni algoritmiche, ba¬ sate sulla stessa sequenza di modelli {yk} (k = 1, ..., R), sono state proposte in letteratura. In particolare è risultata piuttosto efficiente la soluzione proposta da Helland (1990) nella quale il K-esimo modello {y *} viene ottenuto attraverso una regressione a minimi quadrati della variabile di risposta y sulle K combinazioni lineari.

Scìppacercola S. et al. - Il Partial Least Squares

115

Appendice 2

Principal component regression (Massey, 1965)

La Principal Component Regression (PCR) prende le mosse dalla decomposizione in valori singolari della matrice di varianza e covarianza delle variabili predittrici

v = s \kvkvk

k= 1

dove X\ ( k = sono gli auto valori della matrice V ordinati in

maniera non crescente e \k (k = 1, p) i corrispondenti autovettori.

Indicato con R il numero di autovalori non nulli della matrice V, la tecnica proposta da Massey, produce una sequenza di modelli di regres-

sione {y0 ... yR } (K = 1, ..., R )

% = 2 [CyX0vt)At2]X0v, (5)

k= 0

ciascuno ottenuto regredendo y sulle variabili zk = X0vk, dove per K=0 si ha y0 = 0. La metodologia PCR ricerca il migliore modello predittivo yK di rango ridotto con la tecnica dell’Ordinary Cross Validation (CV) (Stone, 1974),

n

K - min 2 (yf - yk/i)2

0 <K<R i= 1

dove yk/i è il k-esimo modello di regressione (5) ottenuto dai dati originali con la i-esima osservazione cancellata.

116

Boll. Soc. Natur. Napoli - Voi. 102 ( 1994-1995 )

m

Appendice 3

Ridge regressione (Hoerl-Kennard, 1970)

L’obiettivo principale della Ridge Regressione (RR) è quello ottenere dei coefficienti stabili in presenza di multicollinearità. I coefficienti a- = (a;i, djp), per / = 1 , q, del modello lineare y] = X0a 7- sono ottenuti con il metodo dei minimi quadrati vincolati, con vincolo proporzionale alla norma quadratica del vettore dei coefficienti a:

àx = min [Z(y - X0a)2 + Aa'a] = (V + A l)~l Xy

a

dove I è una matrice identità di ordine p.

La matrice di varianza e covarianza delle variabili predittrici V mal¬ condizionata è allora stabilizzata aggiungendo a V un multiplo della ma¬ trice identità /, con un grado di stabilità regolato dal valore assunto dal parametro "ridge” A > 0, quest’ultimo stimabile, per esempio, con la tecnica dell’Ordinary Cross Validation. Un valore A - oc si ha nel caso di un modello y = 0 mentre A = 0 porta ad una stima a minimi quadrati ordinaria malcondizionata.

Ringraziamenti. Si ringrazia il Doti. E.