S U M M A RY
1. Structure and diversity of the macroinvertebrate fauna were studied in relation to
altitude and latitude among three groups of streams from Ecuador (lowland: 100–600 m,
Central Valley: 2600–3100 m, páramo: 3500–4000 m), and one group from the temperate
lowland region of Denmark. The streams in the four regions were comparable with
regard to physical characteristics such as size, current and substratum.
2. In terms of faunal composition the Ecuadorian highland streams bore more
resemblance to the Danish lowland streams than the Ecuadorian lowland streams. The
greater similarity between the Ecuadorian highland and the Danish streams, however,
was due to the large number of insect families in the Ecuadorian lowlands, many of
which were not found in the other regions. Of ten physico-chemical parameters
measured, maximum stream temperature explained by far the most variability in faunal
composition.
3. The number of insect orders and families increased linearly with maximum stream
temperature and therefore decreased with altitude and latitude. A compilation of
literature data on insect richness and maximum water temperature from streams around
the world confirmed this pattern, yielding a common linear relation for both temperate
and tropical streams. This pattern may arise due to a direct temperature effect on
speciation but is probably also related to geological history and the influence of climatic
changes on stream ecosystems. We estimate that small, tropical, lowland streams have,
on average, a two- to fourfold higher species richness than temperate lowland streams.
Introduction
Species richness among most groups of organisms in Benson & Smith, 1986), while others have found
the sea and on the continents increases markedly the same diversity in tropical and temperate regions
from high latitudes towards the tropics (Fischer, 1960; (Patrick, 1964; Arthington, 1990; Flowers, 1991). Our
Rosenzweig, 1995). This pattern is much less clear ability to evaluate this question is limited, however,
in the freshwater biota. Species richness of aquatic because few studies are available from the tropics, all
angiosperms is lower in tropical than in temperate differing in sampling technique, and because know-
regions (Crow, 1993), and richness of molluscs is about ledge on the systematics of tropical freshwater insects
the same (Hubendick, 1962). Apparently only the fish is incomplete.
have reached a far greater species richness in tropical Temperature affects stream invertebrates in a num-
freshwaters (Lowe-McConnell, 1987). The latitudinal ber of ways (Ward & Stanford, 1982; Allan, 1995) and
pattern of aquatic insect diversity is unclear, as some is the most apparent abiotic variable related to latitude
authors have found higher diversity in the tropics as well as to altitude. Studies on altitudinal succession
(Bishop, 1973; Stout & Vandermeer, 1975; Pearson, from tropical (Illies, 1964; Hynes, 1971; Williams &
Hynes, 1971) and temperate regions (Kownacka & extensively cultivated areas. The maximum distance
Kownacki, 1972; Allan, 1975; Ward, 1986) demonstrate between any two of the streams was 25 km. The
clear changes in faunal structure and a decrease in lowland streams were sampled in June 1995. The
diversity with altitude. Illies (1964) suggested that an second Ecuadorian group included streams at altitudes
analogous succession in life-types to that from high between 2600 and 3100 m in the Central Valley of the
to low altitudes should also exist from high to low Andes around the capital city of Quito. The streams
latitudes, and temperature was suggested to be the drain secondary scrub vegetation and extensively cul-
main variable governing this pattern. However, longi- tivated areas. These streams were up to 85 km apart
tudinal zonation studies following the same stream and were sampled in September and October 1994.
from mountain brook to lowland river obscure the The third group of Ecuadorian streams was located in
effect of temperature itself because small streams and the Eastern Cordillera of the high Andes at 3500–
large rivers, even at the same altitude, are two different 4000 m a.s.l. These streams drain páramo, a humid
habitat types with different faunal assemblages grassland vegetation characteristic of the Northern
(Vannote et al., 1980), and stream size itself has an effect Andes. The páramo streams were up to 110 km apart
on species richness (Brönmark et al., 1984; Minshall, and they were sampled from February to May 1995.
Petersen & Nimz, 1985; Jacobsen & Friberg, 1997). Finally, a group of Danish lowland streams at 0–
Comparative studies on the overall faunal structure 100 m a.s.l. located in mid-Jutland, was included. The
and diversity of similar stream types located at differ- streams drain secondary deciduous forest, pasture and
ent altitudes (Ormerod et al., 1994) and latitudes (Stout cultivated areas and were maximally 75 km apart. The
& Vandermeer, 1975) are few. Hence, the specific streams in Denmark were sampled from May to July
influence of temperature on faunal composition along 1993. All streams in Ecuador and Denmark included
altitudinal and latitudinal gradients has remained in the study were first-, second- or third-order streams,
elusive. rather fast-flowing with coarse substratum and with
This study compares quantitative composition and no or very little aquatic macrophyte growth. None of
diversity of the macroinvertebrate fauna from four the streams had significant sources of pollution.
groups of streams. Three groups were from different
altitudes in tropical Ecuador, and one group was from
Physicochemical measurements
the temperate, lowland region of Denmark. All streams
were sampled in the same way, and a consistent level In Ecuador, physicochemical parameters were meas-
of identification to family was applied to all samples. ured on three occasions during the dry season in each
Furthermore, the streams were comparable with of the lowland and Central Valley streams, but only
regard to physical parameters such as size, current once in each of the páramo streams. The highest
and substratum. Temperature was the main physical daytime temperature recorded and mean values of all
parameter differing systematically among the stream other parameters are used in this study. Measurements
groups. The purpose of the study was to examine the in the Danish streams were performed once in March.
general patterns in structure and diversity of the The maximum temperatures, however, originate from
invertebrate fauna across altitudes and latitudes and several years of measurements in each stream. Con-
subsequently relate the patterns to stream temperature. ductivity, normalized to 20 °C, was measured with a
WTW LF96 meter, pH was measured with a WTW
pH96 meter, and alkalinity was measured by means
Materials and methods of gran titration with 0.1 N HCl (Mackereth, Heron &
Localities Talling, 1978). Mean current velocity (in 0.4 3 water
depth) was measured with a Höntzch digital anemo-
The study included three groups of streams in Ecuador meter with a vane wheel sensor at four to eight points
and one group in Denmark. Each group consisted of along three cross-sections. The composition of the
eight streams. The first group in Ecuador included substratum was measured by registering the type of
streams located in the coastal lowlands at altitudes of substratum in 58–113 points along six to eight ran-
100–600 m above sea level (m a.s.l.). These streams domly chosen transects across the stream. The mineral
drain disturbed, tropical, lowland forest, pasture and substratum types were assigned to the following six
© 1997 Blackwell Science Ltd, Freshwater Biology, 38, 247–261
Structure and diversity of stream invertebrates 249
categories: silt, sand, gravel, pebble, cobble and were identified using Roldán (1992) and Merritt &
boulder (Hynes, 1970). Cummins (1996).
(°C) (µS cm–1) (mEq l–1) pH (cm) (l s–1) (cm s–1) silt 1 sand gravel 1 pebble cobble 1 boulder
Coastal lowlands 24.8a 40a 0.37a 6.89a 323a 122a 20a 8a 63a 30a
(22.4–26.3) (18–67) (0.17–0.62) (6.30–7.49) (140–643) (6–394) (4–42) (0–29) (46–72) (0–51)
Central valley 14.6b 192b 1.51b 7.88b 227a 148a 24a 25ac 37b 37ab
(10.7–17.8) (101–317) (0.55–2.79) (6.82–8.90) (108–443) (9–689) (7–43) (0–69) (18–66) (7–62)
Páramo 9.8c 56a 0.63a 7.39ab 213a 106a 23a 10a 35b 55b
(6.4–14.8) (19–119) (0.14–1.50) (6.18–8.25) (123–350) (5–289) (5–53) (0–24) (10–56) (39–81)
Denmark 15.3b 331c 1.55b 7.64b 183a 57a 24a 36bc 54a 9c
(12.0–18.0) (165–450) (0.23–2.53) (6.35–8.31) (50–272) (1–120) (8–39) (18–47) (17–69) (0–20)
Ecuador
Lowland 436a 29.3a 3.38a 0.756a
(114–1040) (24–33) (2.89–3.81) (0.622–0.849)
Valley 1382b 20.9b 1.97b 0.516b
(587–4100) (15–26) (1.52–2.27) (0.337–0.751)
Páramo 653ac 13.5c 1.35c 0.634b
(335–1310) (10–20) (1.02–1.84) (0.520–0.799)
Denmark 750bc 20.0b 2.01b 0.556b
(318–1070) (17–24) (1.59–2.48) (0.386–0.644)
had from four to six of the applied inorganic types of streams was lower (70%) (Table 3). The Crustacea
substrate. Temperature clearly divided the streams (Gammaridae) showed a clear increase with altitude,
into three groups: the Ecuadorian lowland streams from absence in the Ecuadorian lowland, 3% in the
had a mean maximum temperature of 24.8 °C, the Central Valley, to 14% on the páramo. Gammarus was
páramo streams of 9.8 °C, while the Central Valley a very prominent part of the Danish stream fauna.
and Danish streams had similar mean maximum tem- Oligochaeta also increased in proportion from the
peratures of 14.6 and 15.3 °C. All chemical parameters Ecuadorian lowland to the páramo. Turbellaria
differed clearly among the stream groups, reflecting (Planaridae) and Bivalvia (Pisidiidae) were other
the nutrient-poor waters of the Ecuadorian lowland important groups in the páramo streams. The number
and páramo streams relative to the Central Valley and of insect orders decreased with altitude. In the lowland
Danish streams. streams nine orders were found. Odonata comprised
8%, Hemiptera 7%, Megaloptera 0.8% and Lepidoptera
0.5% of the lowland fauna, but, except for a few
Structure of fauna
Odonata in the Central Valley, these four orders were
The density of invertebrates was significantly lower absent in the Central Valley and the páramo streams.
in the Ecuadorian lowland streams compared to the Ephemeroptera dominated the insect fauna in all
Central Valley and the Danish streams while the three regions in Ecuador, but were less numerous in
number of higher taxa (insect families 1 non-insect Denmark (Table 3). Plecopterans, in contrast, were rare
classes) was higher in the lowland streams (Table 2). in all three regions in Ecuador but were numerous in
Two indices were calculated to evaluate the two basic the Danish streams. Coleoptera and Trichoptera were
components of diversity, richness and equitability: the more important and Diptera less dominant in the
Margalef richness index as a measure of the number Ecuadorian lowland streams than in the other
of taxa taking the sample size into account, and the stream groups.
Pielou evenness index as a measure of equitability A total of fifty-four families of insects were found
irrespective of the number of taxa. Both the Margalef in Ecuador and forty-four in the lowlands (Table 4).
and the Pielou indices were significantly higher in the Twenty-three insect families were restricted to the
lowland streams than in any other of the three stream lowlands, mainly families of Odonata and Hemiptera,
groups (Table 2). Thus, the fauna of the lowland whereas eleven families were restricted to the
streams were less dominated by few taxa. Central Valley and the páramo. Only one family
The fauna of the lowland and the Central Valley (Limnephilidae: Trichoptera) was restricted to the
streams of Ecuador was completely dominated by páramo. Fifteen families were found at all altitudes
insects (95% and 93%, respectively), while the propor- and many of these were common or abundant at all
tion of insects in the páramo and in the Danish altitudes (Baetidae, Elmidae, Glossosomatidae,
© 1997 Blackwell Science Ltd, Freshwater Biology, 38, 247–261
252 D. Jacobsen, R. Schultz and A. Encalada
Table 3 Composition of the macroinvertebrate fauna given as a total of kick-samples from eight streams in three Ecuadorian and
one Danish region
Ecuador
n % n % n % n %
Ecuador
To test whether the relation between stream temper- temperate-arctic and tropical streams at different alti-
ature and insect family richness was generally valid, tudes, and thereby provided a latitudinal as well as
we compiled data available from the literature on altitudinal gradient. In the plot we used only single
© 1997 Blackwell Science Ltd, Freshwater Biology, 38, 247–261
254 D. Jacobsen, R. Schultz and A. Encalada
Table 5 Data on diversity of aquatic insect families from
Ecuadorian and Danish streams showing α-diversity,
β-diversity, γ-diversity and proportion of the regions family
pool occurring in one stream (α/γ)
α γ β α/γ r2
Ecuador
Lowland 26.1 (22–30) 47.7 0.89 0.55 0.985
Valley 16.9 (13–21) 35.2 1.13 0.48 0.998
Páramo 10.0 (6–16) 24.6 1.48 0.41 0.998
Denmark 15.0 (11–18) 34.8 1.36 0.43 0.998
Peruvian Andes stream, and Statzner (1975) found the closely related to the fauna of the páramo than to the
transition at 1700 m in a stream in Zaı̈re. These findings lowlands. Hynes (1971), however, studied longitudinal
agree with the findings in the present study, as the succession in a stream in Trinidad and found a major
fauna of the Central Valley (2600–3100 m) was more shift in faunal composition at an altitude of only 30 m.
Hynes argued that the transition from rhithron to
potamon fauna occurs where the hill stream meets the
plain and therefore depends on the nature of the
stream bed, and that temperature (which did not vary
much in his study) is yet another factor superimposed
on the stream system. This disagreement reflects the
difficulty of interpreting studies of longitudinal faunal
succession in relation to altitude and suggests that
it may be more appropriate to regard altitudinal
succession in streams as a continuous process rather
than dividing streams into distinct zones. Ward (1985)
has a more thorough discussion of stream zonation in
relation to temperature.
Surprisingly few studies have adressed directly the
importance of temperature as a structuring parameter
for the stream invertebrate communities under normal
conditions in comparable streams. The streams
included in this study did not differ greatly regarding
size, current, substratum or aquatic macrophyte
growth, and the chemical parameters did not explain
much of the faunal variability or family richness.
We suggest therefore that temperature is a major
parameter structuring the stream fauna.
Speciation in relation to temperature and climatic and should also be considered to explain regional
history differences in taxa richness. During the last ice age,
some 20 000 years ago, large lowland areas at temper-
As part of the voluminous debate on the possible
ate latitudes, including most of Denmark, were
causes for latitudinal gradients in biodiversity Rohde
covered by glaciers. In the tropical high Andes the
(1992) concluded that temperature itself may be a main
mean air temperature was about 6 °C lower than
factor governing diversity because high temperature
today, with permanent snow cover probably extending
leads to higher mutation rates and shorter generation
down to 3500–4000 m, and glaciers to an altitude of
times, thereby speeding up evolution and speciation.
3000 m (Colinvaux, 1987; Schubert, 1988). Hence, the
Therefore, more families (and species) of stream insects
highland stream fauna was probably driven down to
have probably evolved in tropical lowland streams
lower altitudes (and the northern fauna to lower
than in tropical mountain or higher latitude streams.
latitudes) and many species became extinct. Con-
It is difficult to say whether temperature during
sequently, most stream faunas of the high Andes and
evolution works on a regional or a local scale. Regional
of northern Europe (or North America) are relatively
(γ) species richness affects, and is tightly related to,
young and still under colonization from lower alti-
local richness (Cornell & Lawton, 1992), obviously by
tudes or more southerly streams.
setting an upper limit to local richness. However,
The neotropical lowlands may also have undergone
temperature was more closely related to γ diversity
climatic fluctuations in historic or geological time
than to local richness, probably because of the con-
leading to extinctions, while still reaching higher
founding effect of local abiotic and biotic conditions
species richness (Fischer, 1960). In particular, lowland
at the individual stream sites. Furthermore, local rich-
streams may have been affected by lower temperatures
ness is probably affected by the partly stochastic
and extensive droughts during the Pleistocene
process of dispersal. Dispersal between localities is
(Covich, 1988). However, such climatic fluctuations
related to the β diversity. The higher β diversity of the
may also have contributed to speciation through isola-
páramo streams relative to the Ecuadorian lowlands
tion in refuges. The northern part of western Ecuador
may simply result from the fact that the páramo
is actually part of a postulated pleistocene refuge
streams covered a larger land area than the Ecuadorian
(Colinvaux, 1987).
lowland streams. However, β diversity may also have
been affected by low dispersal of flying adult insects
The influence of predation and food sources on insect
among the younger mountain streams.
diversity
Temperature itself may not be the sole cause of the
relation between richness and temperature. Species richness in tropical vs. temperate streams has
Quaternary climatic history is related to temperature been thought to be affected by several abiotic and
© 1997 Blackwell Science Ltd, Freshwater Biology, 38, 247–261
258 D. Jacobsen, R. Schultz and A. Encalada
biotic factors such as available food sources, habitat The effect of environmental stability on insect diversity
diversity (Covich, 1988) and predation (Stout &
Vandermeer, 1975; Fox, 1977). No indications were Environmental variability and disturbance is regarded
seen in the present study of a higher habitat diversity as a key factor for biotic diversity in stream com-
or lower food diversity in neotropical streams com- munities (Stout & Vandermeer, 1975; Stanford & Ward,
pared to temperate streams as suggested by Covich 1983; Arthington, 1990; Reice, Wissmar & Naiman,
(1988). Flowers (1991) also proposed that food sources 1990). Although some temperate regions are character-
are basically the same in tropical and temperate stre- ized by highly variable stream flow, tropical lowland
ams, namely detritus and algae. streams in general seem more prone to spates than
Fox (1977) suggested that higher proportions of temperate or high altitude streams, and the global
predators in tropical streams may lower invertebrate maximum stream temperature gradient may run some-
what parallel to a ‘discharge unpredictability gradient’.
densities and increase diversity by maintaining com-
Ward & Stanford (1983) proposed that the general
petitively superior prey species at lower densities and
‘intermediate-disturbance hypothesis’ (Connell, 1978)
thereby allowing the existence of less competitive
should be applicable to stream invertebrate communit-
species. It is noteworthy that the two insect orders
ies, so that maximum richness is reached at inter-
that primarily contribute to the higher species richness
mediate levels of disturbance. However, we do not
of aquatic insects in the neotropics (Odonata and
know the exact optimum level of disturbance. There-
Heteroptera) (Table 6) both are predatory, and these
fore, whether the influence of increasing disturbance
groups mainly occur in the lowland streams. The
has either a positive or a negative impact on species
proportion of invertebrate predators (by numbers) is
richness may depend on the location of the streams
higher in Ecuadorian lowland streams (24%) than in
along the ‘disturbance-gradient’. Although intuitively
Danish streams (6%) (Schultz, 1997), and also higher
appealing, the review of Reice et al. (1990) found little
than that found in the Ecuadorian Central Valley
evidence for the hypothesis. ‘Rock-tumbling experi-
streams (9%) (D. Jacobsen & A. Encalada, unpublished
ments’ teach us little about patterns in diversity among
data). A study on the Ecuadorian páramo streams
streams and very few studies have examined species
showed that predator abundances were generally low
richness and quantified overall stability in a set of
(3%), but local dominance of Planarians may increase
streams. However, in such a study, Death &
this figure to 65% (Encalada, 1997). Hence, high densi-
Winterbourn (1994) found that unstable streams had
ties and diversities of invertebrate predators may fewer species than more stable streams. Hence, there
influence invertebrate community structure and may be a negative impact of the suggested latitudinal
diversity in tropical streams. disturbance-gradient on diversity of stream insects
Neotropical lowland streams often have an closer to the equator. Still, we know little about how
abundant and diverse fish fauna (Lowe-McConnell, variability in flow regime affects regional richness and
1975). Some experimental studies in temperate as well speciation of stream insects.
as in tropical streams (Flecker, 1992; Dudgeon, 1991, Tropical streams are, on the other hand, more con-
1993) have found effects of fish predation on inverte- stant than temperate streams in terms of temperature
brate community structure, but it is unclear just how regime. It has been suggested that a wide annual
important and general the impact from predatory temperature range (as found at temperate latitudes)
fish is. Flecker (1992) found that herbivorous and may enhance species diversity by allowing the coexist-
detritivorous fish, which are much more abundant in ence of differently synchronized life cycles, thereby
neotropical streams than in temperate streams, had a ensuring temporal separation of major periods of
stronger indirect effect than predatory fish on the resource use and reducing active competition (Vannote
dominant invertebrate taxa by depleting food et al., 1980). Likewise, wider daily variation in temper-
resources that were shared between the fish and ature increases species packing by providing a wider
the invertebrates. Both predatory and herbivorous/ range of temperature optima, even though suboptimal
detritivorous fish may thus be an important controlling conditions consequently will occur for each species as
factor for invertebrate community structure and well (see references in Ward & Stanford, 1982). It
diversity in neotropical streams. seems, therefore, that both temperature variability and
© 1997 Blackwell Science Ltd, Freshwater Biology, 38, 247–261
Structure and diversity of stream invertebrates 259
discharge variability should favour higher diversity Macroinvertebrate community structure and
in temperate streams than in tropical streams. environmental characteristics along a large river:
In conclusion, the global relationship between family congruity of patterns for identification to species or
richness and maximum stream temperature is sug- family. Journal of the North American Benthological Society,
15, 232–253.
gested to be related to geological history and climatic
Brönmark C., Herrmann J., Malmqvist B., Otto C. &
changes in addition to the direct effect of temperature.
Sjöström P. (1984) Animal community structure as a
We propose that more families and species have
function of stream size. Hydrobiologia, 112, 73–79.
evolved in warm than in cold regions because of a Clarke K.R. & Warwick R.M. (1994) Change in marine
higher speciation rate for a longer time period (high communities: An approach to statistical analysis and
geological age), and that cold streams at high altitudes interpretation. Plymouth Marine Laboratory. National
in the tropics (the páramo) or at high latitudes Environmental Research Council, U.K.
(Denmark) have young biotas because of former glaci- Colinvaux P. (1987) Amazon diversity in light of the
ations and therefore are still under the process of paleoecological record. Quaternary Science Reviews, 6,
colonization. 93–114.
Colwell R.K. & Coddington J.A. (1995) Estimating
terrestrial biodiversity through extrapolation.
Acknowledgements Biodiversity: Measurement and Estimation (ed. D. L.
Hawksworth), pp. 101–118. Chapman & Hall, London.
We thank Nikolai Friberg for the Danish invertebrate Connell J.H. (1978) Diversity in tropical rainforests and
data, Klaus P. Brodersen for giving us access to the coral reefs. Science, 199, 1302–1310.
PRIMER software, Peter C. Dall for making the jack- Corbet P.S. (1980) Biology of Odonata. Annual Review of
knife computer program and Kaj Sand-Jensen, Niels Entomology, 25, 189–217.
Peder Kristensen and two anonymous referees for Cornell H.V. & Lawton J.H. (1992) Species interactions,
valuable comments on the manuscript. The local and regional processes, and limits to the richness
Departamento de Ciencias Biológicas at the Pontificia of ecological communities: a theoretical perspective.
Universidad Católica del Ecuador in Quito kindly Journal of Animal Ecology, 61, 1–12.
Covich A.P. (1988) Geographical and historical
provided laboratory facilities. This work was funded
comparisons of neotropical streams: biotic diversity
by grant 104.Dan8/630 from Danida, Danish Ministry
and detrital processing in highly variable habitats.
of Foreign Affairs.
Journal of the North American Benthological Society, 7,
361–386.
Crow G.E. (1993) Species diversity in aquatic
References
angiosperms: latitudinal patterns. Aquatic Botany, 44,
Allan J.D. (1975) Faunal replacement and longitudinal 229–258.
zonation in an alpine stream. Verhandlungen der Death R.G. & Winterbourn M.J. (1994) Environmental
Internationale Vereinigung für Theoretische und stability and community persistence: a multivariate
Angewandte Limnologie, 19, 1646–1652. perspective. Journal of the North American Benthological
Allan J.D. (1995) Stream Ecology, Structure and Function of Society, 13, 125–139.
Running Waters. Chapman & Hall, London. Dudgeon D. (1989) The influence of riparian vegetation
Andersen T., Hansen H.O., Jacobsen D., Krøjgaard L. & on the functional organization of four Hong Kong
Poulsen N. (1989) Ørreders (Salmo trutta L.) indvirkning stream communities. Hydrobiologia, 179, 183–194.
på invertebratfaunaen i vandløb, med særlig vægt lagt på Dudgeon D. (1991) An experimental study of the effects
amphipoden Gammarus pulex L. Masters dissertation. of predatory fish on macroinvertebrates in a Hong
Freshwater Biological Laboratory, University of Kong stream. Freshwater Biology, 25, 321–330.
Copenhagen. (In Danish.) Dudgeon D. (1993) The effects of spate-induced
Arthington A.H. (1990) Latitudinal gradients in insect disturbance, predation and environmental complexity
species richness of Australian lotic systems: a selective on macroinvertebrates in a tropical stream. Freshwater
review. Tropical Freshwater Biology, 2, 179–196. Biology, 30, 189–197.
Bishop J.E. (1973) Limnology of a small Malayan river, Elgmork K. & Sæther O.E. (1970) Distribution of
Sungai Gombak. Monographiae Biologicae, Vol. 22. Dr invertebrates in a high mountain brook in the Colorado
Junk, The Hague. Rocky mountains. University of Colorado Studies, 31,
Bournaud M., Cellot B., Richoux P. & Berrahou A. (1996) 1–55.