Definition and process-based classification of caves

Abstract

Cave is re-defined in order to be linked to the cave formation processes, to cover the known cave types, to differentiate from porosity and contiguous spaces, to be applied also in a continuum of size and to avoid explorational bias. Despite the scientific basis, the proposed definition remains simple enough to be used by cavers and non-specialists. Following this definition, a classification scheme that is also process-based combines the known cave types. Clustering is based on five levels of classification, from which the first two levels define the major cave categories. The rest of the branching is the result of variation in settings and formation agents. A discussion on various classifications and definitions reveals the non-static character of such schemes that tend to change in relation to the progress of research cave census and improved communication of scientists on previously and new discovered caves.

Full text

DEFINITION AND PROCESSBASED CLASSIFICATION
OF CAVES
OPREDELITEV IN KLASIFIKACIJA JAM
NA PODLAGI PROCESOV
Georgios LAZARIDIS1
Abstract UDC 551.435.84:001.4
Georgios Lazaridis: Denition and process-based classicati-
on of caves
A new denition of the term “cave” enables its linking to recog-
nised cave formation processes, its coverage of the known cave
types, its dierentiation from porosity and contiguous spaces,
and its applicability within a continuum of sizes, as well as en-
suring avoidance of explorational bias. Despite its scientic ba-
sis, the proposed denition remains straightforward enough to
be used by cavers and by non-specialists. Guided by this de-
nition, a proposed hierarchical classication scheme, which is
also process-based, combines the known cave types. e hier-
archy is based upon ve levels of classication, wherein the rst
two levels dene the major cave categories. Further branching
encompasses variation in cave development settings and in
agents of cave formation. Discussion of various pre-existing
classications and denitions reveals the non-static nature of
such schemes, which tend to evolve in response to the progress
of cave research. e key is in increases in cave census data and
improved communication by and between scientists about pre-
viously and newly discovered caves.
Keywords: cave denition, cave classication, speleogenesis.
Izvleček UDK 551.435.84:001.4
Georgios Lazaridis: Opredelitev in klasikacija jam na pod-
lagi procesov
Nova opredelitev pojma jama omogoča povezavo s prepozna-
nimi procesi nastajanja jam, to, da zajema znane jamske tipe,
da se razlikujejo na podlagi poroznosti in sosednjih prosto-
rov ter uporabnost v okviru proučevanja velikosti, poleg tega
preprečuje pristranskost v raziskovanju. Kljub znanstveni pod-
lagi je predlagana opredelitev dovolj preprosta, da je uporabna
tako jamarjem kot nestrokovnjakom. Na podlagi te opredel-
itve predlagana hierarhična klasikacijska shema, ki prav tako
temelji na procesu, združuje znane tipe jam. Hierarhija temelji
na petih ravneh klasikacije, pri čemer prvi dve ravni opre-
deljujeta glavne kategorije jam. Nadaljnje razvrstitve temeljijo
na razlikah v okolju nastanka jam in dejavnikih nastanka jam.
Razprava o različnih že uveljavljenih klasikacijah in opredelit-
vah razkriva nestatičnost takšnih shem, ki se razvijajo glede na
napredek pri raziskovanju jam. Ključni so večji nabor podatkov
o jamah, sporočanje znanstvenikov in izboljšana komunikacija
med znanstveniki o predhodno in na novo odkritih jamah.
Ključne besede: denicija jame, klasikacija jam, speleogeneza.
ACTA CARSOLOGICA 51/1, 65-77, POSTOJNA 2022
1 Aristotle University of essaloniki, School of Geology, 54124, essaloniki, Greece, e-mail, ORCID: geolaz@geo.auth.gr,
https://orcid.org/0000-0002-4926-2357
Received/Prejeto: 8.3.2022
COBISS: 1.01, DOI 10.3986/ac.v51i1.10611
CC BYNCND

GEORGIOS LAZARIDIS
1. INTRODUCTION
Whereas use of the term “cave studies” is understood in-
ternationally, underlying use and understanding of the
term “cave” continues to hinge upon a conventional but
arbitrary limiting constraint of human size. For example,
according to the most-used denition of the International
Union of Speleology (UIS, Union Internationale de Spé-
léologie), a cave is considered as being “… any natural,
underground cavity, large enough to be entered by man”
(Bögli, 1980). is denition is not genetic and uses (ar-
bitrary) human-size as a measure of caves. It is broadly
applied because most related information comes only
from accessible parts of caves (Ford & Williams, 2007).
However, openings of smaller size may be parts of what
is termed cave and share the same characteristics as they
are both formed under the same formation conditions.
is is the reason for the existence of more specic de-
nitions for some process-based groups of caves, such as
karst (in particular hypergene) caves, where, for example,
Ford and Williams (2007) dene cave as “an opening en-
larged by dissolution to a diameter sucient for ‘break-
through’ kinetic rates to apply if the hydrodynamic setting
will permit them”. In other scientic elds such as biology
various cave denitions are adopted. For example, White
and Culver (2005) dene cave as “a cavity, at least part of
which is in constant darkness, with turbulent water ow
and with eyeless, depigmented species present”. An ecologi-
cal denition of caves is the following: “A cave is a natural
or articial cavity in rock in which large-scale scalar phe-
nomena are actually or potentially ecologically signicant.
ese phenomena include the presence of surfaces (which
may be at rock-air, rock-water and/or water-air interfaces)
available for utilization by mesofauna and/or macrofauna.
Usually, though not invariably, they also include the pres-
ence and eects of uid ow (air currents, streams, springs
or tidal ow) and they also commonly have accumulations
of bulk substrates such as guano, vegetable debris, talus and
sediments. ere is also potential for access and utilization
by ying animals (bats, birds and insects) and by terrestrial
and aquatic animals that are unable, because of their size,
to utilize mesocaverns or smaller voids.” (Moseley, 2009).
Furthermore, a denition introduced by White (1988)
successfully emphasizes how subjective it could be to de-
ne “cave” and the arbitrary use of human-size; “a natural
opening in the Earth, large enough to admit a human being,
and which some human beings choose to call a cave.”Vari -
ous legal denitions of "a cave" for monitoring purposes
are provided by Weigand et al. (2022).
Mylroie (2019) provides the most recent discus-
sion of the subject and addresses several issues around
the current denition of caves, such as what is a cave and
what is in and around of that; temporal and spatial as-
pects; and classication. Most importantly he points out
that it is time for cave scientists better to communicate
the “cave” term with those who have no direct associa-
tion with caves. e potential opportunity to study ex-
traterrestrial caves is the motivation for meeting such a
need. He denes a cave by using an answering process of
the following ve questions “1. How did the void form?
2. How big is it? 3. How long has it lasted? 4. What does it
contain? 5. How does it connect to exterior space?”. ese
have to be answered in order to access the function of the
cave to the extraterrestrial landscape and how life may
have utilized the cave (Mylroie, 2019). Beyond this he
discusses various problems and prerequisites.
e rst question to be answered relates to the
process(es) of cave formation, widely known as speleo-
genesis. is parameter is used in this paper to dene
“cave” and classify processes and consequently caves ac-
cording to studies that exist in the various speleogenetic
disciplines.
Motivation to propose the new classication scheme
presented here was provided by the need to communi-
cate the subject to speleology students via an analytical
classication that would introduce them to cave research.
Available information is combined into a classication
scheme of various levels. Cave denition and cave clas-
sication are presented and discussed in the next two
sections.
2. METHODS
Criteria such as speleogenesis, size, location and content
are considered for the denition of “cave”. To evaluate the
potential advantages of the proposed denition various
specications introduced by Curl (1964), are used. It is
compared with widely used denitions that already exist
(Table 1 and in text references).
Speleogenesis is considered the classifying factor for
the classication scheme. Classic and recent works that
identify and describe cave type variations in dierent
setting and lithology are used (e.g., Ford & Ewers, 1978;
Bögli, 1980; Audra & Palmer, 2011; Kempe, 2012; Bella &
Gaál, 2013; Klimchouk, 2017; Oberender & Plan, 2018;
ACTA CARSOLOGICA 51/1 – 2022
66

DEFINITION AND PROCESSBASED CLASSIFICATION OF CAVES
Gradziński et al., 2018; Mylroie, 2019). Cave types that
share common speleogenesis are grouped together. e
groups are based on the predominant process of speleo-
genesis and composite categories are introduced in order
to include exceptions regarding this criterion. e results
are organized in dierent levels of classication in tree
diagrams. Categories become more detailed as the clas-
sication level increases.
3. DEFINING CAVE
As mentioned above, many studies use human-size
as a limiting metric for caves, commonly by excluding
any smaller voids (e.g., Davies, 1960). e work of Curl
(1964) is especially signicant for two reasons. First, he
describes the cave as a space, not as an object. is impor-
tant distinction emphasizes the need to dene its bound-
aries. Secondly, he recognized the various problems that
result when human-size is used to help constrain cave
denitions. He solved this problem by introducing (and
dening) a human-sized module on the basis of which
caves can be dened. Applying this module, he dened
“proper caves”. He also discussed the entrances as bound-
aries, and the cave content that can bound or ll the
space, depending upon the perspective of the researcher.
Before providing the denition proposed here, it is ap-
propriate to examine the questions set for the exploration
of caves in space (Mylroie, 2019). e rst question for
a scientist concerns the formation process. Subsequent
questions relate to sizes, the age, the content, and its con-
nection to the exterior. us, following this aspect, any
denition should incorporate this information. Another
issue about denitions is that when more specic ones
are proposed (e.g., for karst caves), they cannot be ap-
plied to other process-based cave types.
Attempting to provide a denition that will include
most, if not all, possible cases, the following new deni-
tion is formulated:
Cave is called every non-articial potentially empty
underground space in solid matter that can be formed by
constructional and destructional geological and biological
processes (such as corrosion, erosion/weathering, deposi-
tion, tectonics, mass movement, deformation, animal ac-
tivities or a combination of them) if the same process is
capable of creating openings large enough to be entered by
humans.
Caves are not necessarily air-lled; they can be lled
partially or totally with matter in any state. e cave-
forming process is called speleogenesis. A discussion of
the denition and possible issues follow.
Most of the proposed denitions consider caves as
naturally formed and exclude human-made structures.
is is not absolute. Curl (1964) suggests that this is op-
tional. Parise et al., (2013) and Mylroie (2019) include
articial cavities within the category caves. e denition
proposed above uses the expression “non-articial” to
exclude any voids created by humans. Such voids could
be tunnels, basements, mines, tombs, etc., which could
have conditions, such as climate, similar to the cave en-
vironment, or include processes common in caves, such
as speleothem deposition. Most articial types of under-
ground voids, such as tunnels, tombs and basements, are
commonly not called a cave, except of mines. us, they
are not included in the current denition.
Subsequently the cave is described as an “under-
ground potentially empty space in solid matter”. is part
of the denition locates the cave below surface and most
importantly denes the limits of the void. e space is
dened as potentially empty because it can be found to-
tally or partially lled with any state of matter (i.e., sedi-
ments, water, air or other gases, lava, etc.). e formation
is dened as being in solid matter, to avoid the inclusion
of void formation in liquid matter. Cases of voids in the
liquid state are described as bubbles. Only if the liquid
changes to the solid state might a space be formed that
can be called a cave instead of a bubble. For example, af-
ter solidication, original bubbles in lava become caves
in volcanic rock. In the gaseous state, voids cannot be
formed at all.
e known processes that support formation of
such potentially empty spaces are summarized as con-
structional or destructional geological and biological
processes. is prerequisite makes the denition broad
and genetic. In contrast to previous genetic denitions,
the inclusion of all the processes (excluding human activ-
ity) in this denition allows the term “cave” to be applied
within each of the various process-based groups, i.e., not
only to karst caves. is consideration enhances the de-
scriptive utility of the denition and allows its use across
a spectrum of speleogenetic studies.
e qualifying phrase “if the same process is capable
of creating openings large enough to be entered by humans”
is added as the key factor to dierentiate caves from po-
rosity and contiguous spaces. Essentially it allows that
potentially empty spaces be regarded as caves if they are
formed by the same process or processes that elsewhere
form potentially empty spaces large enough to be entered
ACTA CARSOLOGICA 51/1 – 2022 67

by humans. us, although a human module remains as
an intrinsic part of the denition, it nally conforms to
the idea of Curl (1964) and embraces caves along a con-
tinuum of sizes. is renement allows the denition to
be used legitimately in deriving statistics of cave evolu-
tion and speleogenesis.
Another crucial issue related to cave denitions is
the question of explorational bias (Mylroie, 2019). For
example, is a lava tube transmitting uid lava a cave or
not? According to the proposed denition, this can be re-
garded as a cave because, potentially, the same processes
can creates empty spaces in the solid state. With regard
to entrances, this denition, as opposed to the classical
ones, also allows inclusion of entrance-less caves, again
potentially removing some of the scope for explorational
bias.
A schematic denition of caves is provided in Figure
1, which takes into account the issues discussed.
Several specications for cave denitions (Curl,
1964) are summarized in Table 1. e potential advan-
GEORGIOS LAZARIDIS
Table 1: Specications for various cave denitions, introduced by Curl (1964) and presented with modications and additions. e term
‘proper’ corresponds to human size as a module.
Limits Explorer Module Entrances Purpose
present work solid state human proper or smaller not required see text
Ford & Williams, 2007 soluble rock water dimensions for
turbulent ow unspecied speleogenesis
White & Culver, 2005 not specied but
apparently rock size of leaving
organisms dimensions for
turbulent ow unspecied biological
research
White, 1988 rock human proper unspecied descripve
UIS, Bögli 1980; Bates &
Jackson, 1987 rock human proper proper descripve
Woodward, 1961 rock and ll water freely owing water not required speleogenesis
Curl, 1960 rock, ll and water human proper or smaller proper stascal theory
of evoluon
Howard, 1960 solid rock unspecied apparently small not required descripve
Davies, 1960; 1964 soluon surface unspecied larger than “primive
tubes” unspecied speleogenesis
Bretz, 1956 rock and ll human proper unspecied descripve
Cullingford, 1953 rock and ll human proper unspecied descripve
Figure 1: Schematic summary of caves as dened in this paper. White and grey circles represent voids formed by dierent processes. e
colored background represents solid matter. A) Voids related to two processes are represented; the white ones are small inaccessible voids
such as inter-granular porosity, formed by processes that cannot normally create voids large enough for human entry; the grey ones repre-
sent caves that are dened by the human size but also smaller voids that are also caves because they are formed by the same process. In this
way explorational bias and human size limitations are avoided, and the denition is ba sed on the process. B) is gure represents an area
where caves large enough for human entry have not yet been found. e grey circles can nevertheless be described a s caves because they are
formed by the same process(es) that formed the grey circles in gure A. us, these are considered to be caves according to the proposed de-
nition because they are formed by a process (or processes) that generally can create spaces large enough for human entry. Any cave-lling
material that obscures exploration (as illustrated by the grey circles with horizontal lines) is irrelevant, because even though inaccessible to
humans, the lled cavities are formed by a process that creates caves. is consideration also avoids explorational bias.
ACTA CARSOLOGICA 51/1 – 2022
68

tages of the proposed denition are readily recognizable.
In most cases the ‘limits’ of the space (cave) is dened
by the rock and maybe the lling. In the present work
the term ‘solid state’ is used to include all rocks, but also
to acknowledge other possibilities, such as caves existing
within metallic asteroids in space (Mylroie, 2019). As in
almost all the general denitions of cave, the ‘Explorer’
is (necessarily) the average human. In the proposed
denition the ‘Module’, however, is proper or smaller,
in contrast to most denitions that use the arbitrary hu-
man size as a limiting measure. ere are no stipulations
about ‘Entrance’ parameters. In most denitions these
are not specied, or is the entrance is required to be at
least of human size (called a ‘proper entrance‘ by Curl,
1964). e ‘Purpose’ parameter is discussed above within
the analysis of the denition’s parts. In conclusion, it is
intended that the present denition covers the totality of
the usages that previous denitions have included.
4. PROCESSBASED CLASSIFICATION OF CAVES
Variations in cave morphology, location, total area, de-
posits, etc. are results of variation in speleogenetic pro-
cesses and settings, oering the possibility of cave clas-
sication relying on dierent internal characteristics
and external factors (e.g., Ford & Williams, 2007). Some
of these characteristics are the size, the cave pattern in
ground plan, the meso-morphology such as passage ge-
ometry, the hydrological setting of speleogenesis and the
cave deposits. External factors are related to lithology, to-
pography, climate, and geomorphological and hydrologi-
cal cycles. All these classications may be useful for ad-
dressing dierent problems in a variety of cave and karst
studies contexts. Furthermore, some of them can be re-
lated to speleogenesis, whereas others cannot. However,
it is notable that internal and external factors are not nec-
essarily separated from and independent of each other.
Several classication schemes based on numer-
ous criteria can be found among the scientic publica-
tions of the last two centuries (e.g., Virlet, 1835; Kraus,
1894; Trimmel, 1968; Bögli 1980; White & Culver, 2005;
Striebel, 2005; Klimchouk, 2006; Ford & Williams, 2007;
Oberender & Plan, 2013; 2018; Bella & Gaál, 2013; Myl-
roie, 2019). Because a process-based cave denition is
proposed herein, there is a related need to recognize,
gather, and map all the various processes involved in the
formation of caves. e classication scheme presented
in gures 2–4 is assembled by considering the speleogen-
esis as the classifying factor, and combining aspects from
previous publications that relate to various cave types.
Diculties arise when discussing morphologies, because
there are a number of contrasting interpretations among
researchers. For example, the speleogenesis of maze caves
(Palmer, 2000), has been part of a long-lasting discussion.
Caves with multiple loops, included in the second and
third stages of the four-stage model (Ford, 1971; Ford &
Ewers, 1978) are attributed by Audra and Palmer (2011)
to development under epiphreatic conditions. us, re-
ecting the understanding that genetic processes impact
the subsequent cave morphology, speleogenetic criteria
were used. e proposed classication aims to employ
and understanding of speleogenetic processes to inform
the establishment of solid categories that will accommo-
date the various morphologies. In a few exceptions, the
categories may directly be considered morphogenetic
(e.g., pyroducts); these have been included in the classi-
cation scheme (Figures 2–4), because some identiable
variations in the processes might exist.
e proposed classication is driven by distinctions
within the mechanics and dynamics that are involved in
the processes, as in other geomorphological sub-disci-
plines (including uvial, aeolian, glacial, groundwater,
etc.).
Five levels of classication emerged during develop-
ment of the proposed scheme (Figures 2–4) with the aim
of classifying caves along with processes of speleogenesis.
us, the cave groups represent either integrated caves
and cave systems or parts of them. For example, hyper-
gene caves consisted of a vadose part connected to the wa-
ter table, and conduits within the phreatic and epiphreat-
ic zones are scrutinized independently at higher levels of
classication. e rst classication level is based upon
the distinction between constructional caves, when the
space pre-exists and the boundary is formed later, and
destructional caves, where the boundary already exists
and the space is created within (Mylroie, 2019).
e fundamental cave types are included in the sec-
ond classication level, which is based on the main pro-
cesses of cave formation. Although, the predominant pro-
cess denes the cave groups of the classication scheme,
it is worth mentioning that multiple processes can act
simultaneously. If a dominant process cannot be recog-
nized, they are grouped as “composite caves” (Figure 2).
e second level of classication is further divided into
subgroups due to variations in the conditions and setting
of the formational processes. e classication scheme is
DEFINITION AND PROCESSBASED CLASSIFICATION OF CAVES
ACTA CARSOLOGICA 51/1 – 2022 69

paired with short descriptions and some comments that
are intended to clarify the distinction followed.
A. CONSTRUCTIONAL CAVES FIGURES 2 & 3
ese caves are formed concurrently with the formation
of the host rock.
1. Synsedimentary caves: they are formed in clastic and
chemical sediments by depositional processes.
a. Progradational caves: they are formed by the pro-
gradation of the steeply sloping surfaces of traver-
tine terraces/masses (Gradziński et al., 2018). As
dened by Pentecost (2005), no distinction is made
between travertine and tufa. ey are further di-
vided into the most common type of caves, formed
in waterfall sites, and those that are developed as
travertine bridges in narrow valleys when several
prerequisites are met (Bayari, 2002).
b. Aggradational caves: these are formed by the aggra-
dation of travertine in artesian springs (Gradziński
et al., 2018).
c. Talus caves: this type is found when caves are formed
among large boulders. e boulders may originate
in several ways (Bella & Gaál, 2013) and the caves
can be further divided, mainly into morphological
subtypes (Halliday, 2006a) that are not described
separately here.
GEORGIOS LAZARIDIS
Figure 2: Division of cave form-
ing processes into three major cave
groups. Constructional, destruc-
tional and multi-process caves.
Groups are explained in the text.
Figure 3: Classication of constructional caves. e second level represents the major groups, and higher levels are due to variations in
conditions and settings.
ACTA CARSOLOGICA 51/1 – 2022
70

d. Imprints: caves formed when travertine or lava
encloses an organism that disintegrates over time,
leaving an empty space. Tree trunks are a common
example (Gradziński et al., 2018). Removal time of
the organic matter diers in such cases but accord-
ing to the proposed denition both types are caves,
because potentially empty space is created with the
deposition of lava or travertine. However, it is a
good example of the temporal aspects as discussed
by Mylroie (2019).
2. Biogenic caves: these are formed by organisms such as
coral-reef builders (e.g., Trimmel, 1968; Bögli, 1980).
Biogenic caves can also be destructional as mentioned
bel ow.
3. Volcanic caves: they are developed in rocks originating
from low-viscosity lavas due to factors including lava
ow volumes and velocities, unequal cooling, degassing
and deformation by lateral forces (Kempe, 2012).
a. Pyroducts: caves formed by owing lava, either due
to ination of older beds or by crust formation of the
outer bed. Both processes form similar morpholo-
gies that are further dierentiated morphologically
and possibly genetically into three categories of in-
creasing complexity: single-; double- or multiple-
trunked and superimposed-trunked systems.
b. Vents: mainly sub-vertical caves in volcanoes that
are formed when the lava vent is not relled.
c. Hollow tumuli: caves formed inside low-prole hills
in the volcanic-ow landscape (tumuli) due to still-
uid lava draining away from inside the mounds.
d. Pressure ridge caves: low and wide caves that are
formed by lateral pressure of solidied lava beds
while the underlying bed is moving.
e. Partings: these are formed when vesicles are devel-
oped due to degassing during lava cooling.
B. DESTRUCTIONAL CAVES FIGURES 2 & 4
ese caves are formed in a pre-existing host rock.
1. Weathering/erosion caves: in this category weather-
ing and erosion are the dominant cave formation pro-
cesses. Various processes and lithologies are involved.
Nine subtypes are recognized:
a. Wave-cut caves: these are formed by the erosional
action of waves on the host rock.
b. Fluvial caves: their formation is related to the ef-
fects of uvial erosion. ree subtypes are included:
riverbank erosion caves by laterally directed uvial
erosion; waterfall erosion caves by backward-direct-
ed erosion of bedrock below and behind waterfalls;
and uvial channel erosion caves that are formed by
erosion cutting into the channel oor (Bögli, 1980;
Kempe & Werner, 2003; Bella & Gaál, 2013).
c. Eolian caves: these are formed by abrasive erosion
related to winds.
d. Suosional/piping caves: open spaces are formed
by the slow or catastrophic removal of matrix and
clasts due to seepage and waterow.
e. Frost weathering caves: processes of rock breakage
related to freezing conditions are responsible for
their formation (Oberender & Plan, 2015).
f. Salt weathering caves: they are formed by rock dis-
integration related to intergranular salt crystalliza-
tion.
g. Mudow caves: these are formed on slopes of mud
volcanoes due to mud outow between dried indu-
rate crust (Bella & Gaál, 2013).
h. Exfoliation caves: caves formed along ssures due to
exfoliation of rocks.
i. Tree moulds: this type includes cavities created by
mechanical removal of petried wood buried in
sediments (Bella & Gaál, 2007).
2. Karst caves: the main cave-forming agent is rock dis-
solution.
a. Hypergene caves: these caves are formed by me-
teoric water that generally moves downwards and
laterally towards a spring or spings. e term hy-
pergene has been proposed by Dublyansky (2014),
better to describe what has commonly been called
"epigene” in speleological studies, for a number of
reasons well established by the author. Hypergene
caves are divided into categories according to the
four-stage model of Ford and Ewers (Ford & Ew-
ers 1978; Ford & Williams, 2007) and the model of
Audra and Palmer (2011), who also used the term
“per ascendum”, which is restricted to development
of hypergene caves related to “water-table rise”. e
term “per descendum” is used here as the opposite of
per ascendum, and it relates to caves/passages devel-
oped as a result of “water-table drop”. e “epiphre-
atic caves with loops” group is also added according
to the model of Audra and Palmer (2011) and in-
terprets dierently part of the four-state model. e
rest groups and their interpretations can be found in
both models.
i. Vadose caves: they are developed in the vadose
hydrological zone where water moves down-
wards; include three subtypes: the basic vadose
caves that are formed in rocks when the water ta-
ble is initially deep; the drawdown caves that are
formed in rocks with initially shallow water table
which drops down as breakthrough advances; in-
vasion caves formed by streams that invade pre-
existing drawdown systems.
ii. Phreatic caves: they are formed in the phreatic
hydrologic zone.
DEFINITION AND PROCESSBASED CLASSIFICATION OF CAVES
ACTA CARSOLOGICA 51/1 – 2022 71

iii. Epiphreatic caves with loops: formed in the epi-
phreatic zone.
iv. Base level caves: they are formed along the water
table.
v. Multistage systems: these are composite cave
systems with complex evolutionary history that
result in the occurrence and succession of several
processes that create the above-mentioned hy-
pergene cave types. ey are divided into those
that follow a rising or dropping base-level; per
ascendum and per descendum speleogenesis,
respectively. It is worth noting that both terms
are dened by base-level changes and not by the
direction of water movement (ascending or de-
scending).
b. Hypogene caves: the recharge of these caves comes
from underlying hydrostratigraphic units and is in-
dependent of the adjacent surface; the uids have
a distant, estranged or deep source. In dominant-
ly vertical parts of these systems the overall water
movement is upwards. Classication of hypogene
caves follows Klimchouk (2017).
i. Artesian hypogene caves: they are formed in
conned multi-story aquifer systems by their hy-
draulic communication.
ii. Endogenous hypogene caves: the process is based
on upwelling ow in, and from deep zones of u-
id-geodynamic inuence. Volcanogenic degas-
sing and other non-volcanogenic volatiles (cold
degassing; see Klimchouk, 2017) can inuence
the process and thus, are used as further division.
iii. Combined artesian and endogenous caves: when
uids of deep origin (basinal/basement) ascent
through cross-formational discontinuities can
interact with the regime of artesian hypogene
speleogenesis and this results in the formation of
caves.
iv. Hypogene caves inside open and incised aquifers:
they are formed in a relatively shallow environ-
ment and result in the formation of two cave-
types. Sulfuric acid speleogenesis (SAS) hypo-
gene caves are formed close at the water table by
water rich in hydrogen sulde that is oxidized to
sulfuric acid (Klimchouk, 2017). e second type
are coastal hypogene caves, which are formed in
the mixing zone between fresh water and sea wa-
ter (Klimchouk, 2017; Mylroie & Mylroie, 2017).
3. Mass-movement and deformation caves of mechani-
cal origin.
a. Crevices: these are formed as narrow rectilinear
caves by mass-movement. ey can form single pas-
sages or passage networks (Halliday, 2006b; Self &
Farrant, 2013).
b. Falling-out caves: formed by the displacement or re-
moval of blocks due to gravity.
c. Caves related to volumetric changes: are formed in
some evaporites/diapirs due to the eects of hydra-
tion and deformation (Bella & Gaál, 2013, Gorbu-
nova, 1978; Reinboth, 1997; Calaforra & Pulido-
Bosch, 1999; Kendall & Warren, 1987; Vendeville &
Jackson, 1992).
d. Collapse shas: these are formed due to ceiling col-
lapse of underground caves. Realistically, almost
all cave types discussed here can include collapse
shas. To allow their formation a pre-existing void
is needed below the incipient sha. If that void is
inaccessible and cannot be studied (explorational
bias), the sha cannot be attributed as part of a par-
ticular cave system. us, mass-movement remains
the driving process and that explains the need for
additional classication levels. Otherwise, such
special cases will remain unclassied or included
erroneously in broader categories based on more-
general criteria such as lithology. Even though rec-
ognition of this category allows an explorational
bias to be introduced, there is no bias related to the
specic cave-forming processes that are the basis of
this scheme.
4. Tectogene caves (Bella & Gaál, 2013 and references
therein): these are formed as a result of tectonic activ-
ity and deformation.
a. Fault caves: they are formed in an extensional geo-
dynamic regime along faults and ssures.
b. Fold caves: they are formed due to unequal defor-
mation of adjacent rock beds during the tectonic
activity that produces folds.
5. Pyrogenic caves: these spaces are created by the burn-
ing-out of coal or organic material (i.e., Dubljanskij
& Andrejčuk, 1989; Bella & Gaál, 2013). It is notable
that the development process corresponds to chemi-
cal removal, and they should not be confused with the
pyroducts mentioned above in the volcanic caves sec-
tion.
6. Ghost-rock karstication: this type of cave is formed
when various types of altered rock are developed lo-
cally within a rock bed or succession during early
stages of diagenesis. Caves may then be formed by
later removal of susceptible material from the zone of
altered rock (Quinif, 2010; Dubois et al., 2014).
7. Glacier caves: formed by the melting of ice and the re-
lated “erosional” eects of meltwater in glaciers.
8. Magmatic caves: these geode-like cavities of various
sizes are most commonly found in plutonic rocks
(Dubljanskij & Andrejčuk, 1989; Bella & Gaál, 2013).
9. Biogenic caves: these are voids that are excavated by
animals (e.g., Lundquist & Varnedoe, 2006).
GEORGIOS LAZARIDIS
ACTA CARSOLOGICA 51/1 – 2022
72

C. MULTIPROCESS CAVES Figure 2
is group accommodates caves that owe their origins to
multiple processes.
1. Composite caves: two or more processes acted simul-
taneously to develop two or more caves that are subse-
quently interconnected.
2. Overprinted caves: pre-existing caves of a specic type
are aected and transformed by the action of process-
es diering from those that formed the original voids.
Depending on which question is addressed, all the clas-
sications available in the literature can be used at least
in part. Some of the main dierences between them are
discussed below.
e classication of Bögli (1980) denes primary
and secondary caves, following earlier works by Kraus
(1894;as cited by Oberender & Plan, 2018 and Trimmel,
1968). Subdivisions of exogenous and endogenous types
are recognized in the secondary caves. eir division
depends upon the dominant cave-forming agent. is
classication can provide information about the speleo-
genesis and the processes involved, but without subtypes.
Furthermore, the scheme does not include categories
that were recognized and dened later, such as hypogene
caves. Despite its broad applicability it was not adopted
by English-speaking researchers until recently (Oberen-
der & Plan, 2018), when caves were classied as con-
structional and destructional by Mylroie (2019). ese
latter terms can be considered synonymous with primary
and secondary, respectively.
e scheme of White and Culver (2005) includes
only the major cave types. Caves developed by dissolu-
tion are further divided by lithology and then by water
chemistry. However, based in some cases on recent ideas,
such as the denition of hypogene caves, hydrological
criteria dominate over geochemical ones. Nevertheless,
water chemistry is also important in speleogenesis and
especially in the case of karst caves. For example, there
are hypogene caves formed by carbonic acid in carbon-
ates, by hydrolysis of gypsum, by sulfuric acid speleogen-
esis, by mixing corrosion, etc. is classication consid-
ers various geochemical controls in the karst speleoge-
netic processes.
e scheme of White (1988) divides caves accord-
ing to chemical and mechanical processes. Klimchouk
(2006) gives the following cave types: solution, volcanic,
glacier, crevice, littoral, piping, and erosion. Both sugges-
tions include only major divisions. Striebel (2005) pro-
posed a classication based on lithology and cave-form-
ing processes (Oberender & Plan, 2018). Palmer (2007)
also used lithology and morphogenetic criteria for classi-
cation. Lithology seems to be signicant for many clas-
sications but there are processes that are not restricted
to specic rock types (Oberender & Plan, 2018).
Mylroie (2019) divides caves into constructional
and destructional. is aspect is also adopted here. He
also includes articial caves in the context. In the pro-
posed classication scheme, it can be observed that bio-
genic caves can be both constructional and destructional
features. Human-made underground voids also t within
these two formation options and one can consider them
part of the wider grouping of biogenic caves, even though
they are excluded from the proposed cave denition.
A classication of non-dissolution caves is provided
by Bella and Gaál (2013); it is a process-based scheme
with 56 subtypes that may be genetic or morphological.
In some cases, cave types, such as boulder caves (e.g.,
glacial, in lava ows, seismotectonic, rockslide boul-
der caves, boulder exfoliation caves) or collapsed caves
(e.g., collapsed pit craters, suosion collapse shas),
are included within several processes. In the proposed
classication scheme, boulder and collapsed caves are
considered as synsedimentary and categorized under
the mass-movement subtype, respectively. Some other
subtypes, such as the tafoni, are considered to be mor-
phological forms, and they are included in the proposed
scheme mainly within the salt weathering group. In addi-
tion, this cave morphotype has been explained by many
processes, which complicate usage of the term. Tectogene
caves are not included among those related to deforma-
tion because they are caused by endogenous forces rather
than the exogenous ones that form mass-movement and
other deformation caves.
Many of the caves classied can be subject to chang-
es by the predominant cave-forming agent. For example,
mature karst cave systems may go through substantial
modications due to erosion (Klimchouk, 2006). In such
cases, a genetic classication may fail to classify it ade-
quately. To remedy this shortfall, the multiprocess group,
as a major type, and multi-stage hypergene subtypes have
been added to the classication scheme within the fourth
level of classication. Many other examples of overprint-
ed processes and composite caves can be recognized.
Ghost-rock karstication is another complex pro-
cess that is dened as a cave group at the rst classication
level. ey are dierentiated from karst and mechanical/
weathering caves because their formation combines as-
pects of both processes in two successive stages of chemi-
cal alteration and material removal. A new perspective
relates both (dissolution and weathering/erosion) with
the endogenous processes of hypogene speleogenesis
(Klimchouk, 2017). Considering all these factors, ghost-
rock karstication is connected provisionally with both
categories in Figure 4.
DEFINITION AND PROCESSBASED CLASSIFICATION OF CAVES
ACTA CARSOLOGICA 51/1 – 2022 73

5. CONCLUSIONS
e new cave denition proposed herein has the main
advantages that:
• it is linked to the cave formation processes,
• it covers the known cave types,
• it uses (typical) human size to dierentiate from po-
rosity and contiguous spaces,
• it applies also in a continuum of sizes even smaller
than human dimensions and
• it is independent of explorational bias.
ese characteristics of the cave denition allow it to be
applied on descriptive purposes, speleogenetic studies,
and statistical analysis. Apart from its scientic ground,
it remains simple enough to be used by cavers and non-
specialists.
e classication scheme is analytical and com-
bines aspects of the most widely used grouping systems
that have been developed to date. Grouping of the basic
processes are encompasses depositional, mechanical and
chemical rock destruction categories. Settings and spe-
cic formational agents provide the additional branches
of the classication.
us, the process-based classication scheme rec-
ognizes 3 main groups with 12 main branches. ese
are the rst two levels of the classication, referencing
GEORGIOS LAZARIDIS
Figure 4: Major groups of destructional caves (2nd level of classication) and their process-based clusters indicate variations of conditions
and setting. e upper right corner is the legend for the classication levels. For group descriptions see text.
ACTA CARSOLOGICA 51/1 – 2022
74

all the known major processes that create caves. e
next three levels of the classication add details that
summarize current knowledge derived from speleo-
genetic studies into a state-of-the-art scheme with 51
endmembers.
A comparison of the various classications that have
been proposed previously in relevant publications reveals
that such schemes are inevitably non-static in character.
ey tend to change in response to the progress of re-
search, more detailed cave census data and improved
communication by and between scientists on aspects of
previously and newly discovered caves.
ACKNOWLEDGMENTS
e author would like to deeply thank the reviewers for
their helpful comments and positive suggestions that
aided improvement of the original manuscript. Dora De-
spoina is thanked for her suggestions on an early version
of the manuscript.
REFERENCES
Audra, P. & A.N. Palmer, 2011: e pattern of caves: con-
trols of epigenic speleogenesis.- Géomorphologie:
relief, processus, environnement, 17, 4, 359–378.
DOI: 10.4000/geomorphologie.9571a
Bates, R.L. & J.A. Jackson, 1987: Glossary of geology.-
American geological institute,pp. 788, Alexandria,
Virginia.
Bayari, C.S., 2002: A rare landform: Yerköprü travertine
bridges in the Taurids Karst Range, Turkey.- Earth
Surface Processes and Landforms, 27, 6, 577–590.
DOI: 10.1002/esp.337
Bella, P. & L. Gaál, 2007: Tree mould caves within the
framework of cave genetic classication.- Nature
Conservation, 63, 7–11.
Bella, P. & L. Gaál, 2013: Genetic types of non-solution
caves.- In: Filippi, M. & P. Bosak (Eds.) 16th Inter-
national Congress of Speleology – Proceedings, 21th–
28th September 2013, Brno, Czech Republic, pp.
237–242, Brno.
Bögli, A., 1980: Karst hydrology and physical speleology.-
Springer Verlag. pp. 284, Berlin.
Bretz, J.H., 1956: Caves of Missouri.- State of Missouri,
Department of Business & Administration, pp. 490.
Calaforra J.M. & A. Pulido-Bosch, 1999: Genesis and
evolution of gypsum tumuli.- Earth Surface Pro-
cesses and Landforms, 24, 10, 919–930. DOI:
10.1002/(SICI)1096-9837(199909)24:10<919::AID-
ESP20>3.0.CO;2-D
Cullingford, C.H.D. (Ed.), 1953: British Caving An Intro-
duction to Speleology.- Routledge. pp. 438, London.
Curl, R.L., 1960: Stochastic models of cavern develop-
ment.- Bulletin of the National Speleological Soci-
ety, 22, 1, 66–76.
Curl, R.L., 1964: On the denition of a cave.- Bulletin of
National Speleological Society, 26, 1, 1−6.
Davies, W.E., 1960: Origin of caves in folded limestones.-
Bulletin of the National Speleological Society, 22, 1,
5−18.
Dublyansky, Y., 2014: Hypogene speleogenesis – dis-
cussion of denitions.- Hypogene cave morpholo-
gies,Karst Waters Institute Special Publication, 18,
1–3.
Dubljanskij, V.N., & V.N., Andrejčuk, 1989: Speleologija
(terminologia, svjazi s drugimi naukami, klassi-
kacija polostej).- Uralskoe otdelenie AN SSSR, Kun-
gur (in Russian).
Dubois, C., Quinif, Y., Baele, J.M., Barriquand, L., Bini,
A., Bruxelles, L., Dandurand, G., Havron, C.,
Kaufmann, O., Lans, B., Maire, R., Martin, J., Ro-
det, J., Rowberry, M.D., Tognin P. & A. Vergari,
2014: e process of ghost-rock karstication and
its role in the formation of cave systems.- Earth-
Science Reviews,131, 116–148. DOI: 10.1016/j.ear-
scirev.2014.01.006
Ford, D.C., 1971: Geologic structure and a new expla-
nation of limestone cavern genesis.- Transactions
of the Cave Research Group of Great Britain, 13,
81–94.
Ford, D.C. & R.O. Ewers, 1978: e development of lime-
stone cave systems in the dimensions of length and
depth.- Canadian Journal of Earth Sciences, 15,
1783–1798.
DEFINITION AND PROCESSBASED CLASSIFICATION OF CAVES
ACTA CARSOLOGICA 51/1 – 2022 75

GEORGIOS LAZARIDIS
Ford, D.C. & P.D. Williams, 2007: Karst hydrogeology and
geomorphology.- John Wiley & Sons, pp. 562.
Gorbunova, K.A., 1978: Peščery gidratacii.- Peščery, 61–
63 (in Russian).
Gradzinski, M., Bella, P. & P. Holubek, 2018: Construc-
tional caves in freshwater limestone: A review of
their origin, classication, signicance and global
occurrence.- Earth-Science Reviews, 185, 179–201.
DOI: 10.1016/j.earscirev.2018.05.018
Halliday, W.R., 2006a: Talus caves.- In: Gunn, J. (Ed.) En-
cyclopedia of Caves and Karst Science. Fitzroy Dear-
born, pp. 1544–1550, New York.
Halliday, W.R., 2006b: Crevice caves.- In: Gunn, J. (Ed.)
Encyclopedia of Caves and Karst Science. Fitzroy
Dearborn, pp. 510–518, New York.
Howard, A.D., 1960: Geology and origin of the crev-
ice caves of the Iowa, Illinois and Wisconsin lead-
zinc district.- Journal of Yale Speleological Society,
2(4,1):5.
Kempe, S., 2012: Volcanic rock caves.- In: White W.B. &
D.C. Culver (Eds.), Encyclopedia of Caves (second
ed.), pp. 865–873, Elsevier.
Kempe, S. & M.S. Werner, 2003: e Kuka’iau Cave,
Mauna Kea, Hawaii, created by water erosion: A
new Hawaiian cave type.- Journal of Cave and Karst
Studies, 65, 1, 53–67.
Kendall C.G.St.C. & J. Warren, 1987: A review of the
origin and setting of tepees and their associated
fabrics.- Sedimentology, 34, 6, 1007–1027. DOI:
10.1111/j.1365-3091.1987.tb00590.x
Klimchouk, A. 2017: Types and Settings of Hypogene
Karst.- In: Klimchouk, A., Palmer, A.N., De Waele,
J., Auler, A.S. & P. Audra (Eds.), Hypogene Karst Re-
gions and Caves of the World, pp. 1–39, Springer.
Klimchouk, A., 2006: Caves.- In: Gunn, J. (Ed.) Encyclo-
pedia of Caves and Karst Science. Fitzroy Dearborn,
pp. 417–421, New York.
Kraus, F., 1894: Höhlenkunde – Facsimile reprint 2009,
Die Höhle, Supplement 56.
Lundquist, C.A. & W.W. Varnedoe Jr., 2006: Salt inges-
tion caves.- International Journal of Speleology, 35,
1, 2. DOI: 10.5038/1827-806X.35.1.2
Moseley, M., 2009: Size matters: scalar phenomena and a
proposal for an ecological denition of ‘cave’.- Cave
and Karst Science, 35, 3, 89–94.
Mylroie, J., 2019: Caves in space.- Journal of Cave & Karst
Studies, 81, 1, 25–32. DOI: 10.4311/2018ES0102
Mylroie, J. & J. Mylroie, 2017: Bahamian ank margin
caves as hypogene caves.- In: Klimchouk, A., Palm-
er, A.N., De Waele, J., Auler, A.S. & P. Audra (Eds.),
Hypogene Karst Regions and Caves of the World,
pp. 757–768, Springer.
Oberender, P. & L. Plan, 2013: A genetic classication
of caves in lower Austria.- In: Filippi, M. & P. Bo-
sak (Eds.) 16th International Congress of Speleology
– Proceedings, 21th–28th September 2013, Brno,
Czech Republic, pp. 390–392, Brno.
Oberender, P. & L. Plan, 2015: Cave development by frost
weathering.- Geomorphology, 229, 73–84. DOI:
10.1016/j.geomorph.2014.07.031
Oberender, P. & L. Plan, 2018: A genetic classication of
caves and its application in eastern Austria.- Geo-
logical Society, London, Special Publications, 466, 1,
121–136. DOI: 10.1144/SP466.21
Palmer A.N., 2000: Hydrogeologic control of cave pat-
terns.- In: Klimchouk A.B., Ford D.C., Palmer A.N.
& W. Dreybrodt (Eds.) Speleogenesis: Evolution of
Karst Aquifers. National Speleological Society, pp.
77–90, Huntsville.
Parise, M., Galeazzi, C., Bixio, R. & M. Dixon, 2013:
Classication of articial cavities: A rst contribu-
tion by the UIS Commission.- In: Filippi, M. & P.
Bosak (Eds.) 16th International Congress of Speleol-
ogy – Proceedings, 21th–28th September 2013, Brno,
Czech Republic, pp. 230–235, Brno.
Pentecost, A., 2005: Travertine.- Springer Verlag, pp. 445,
Berlin.
Quinif, Y., 2010: Fantômes de roche et fantômisation –
Essai sur un nouveau paradigme en karstogenèse.-
Karstologia Mémoires, 18, 57–58.
Reinboth, F., 1997: Die Zwerglöcher bei Walkenried am
SüdharzBemerkumgen zur Frage der Quellung-
shöhlen. Die Höhle, 48, 1, 1–13 (in German).
Self, C., & A.R. Farrant, 2013: Gulls, gull-caves and cam-
bering in the southern Cotswold Hills, England.- In:
Filippi, M. & P. Bosak (Eds.) 16th International Con-
gress of Speleology – Proceedings, 21th–28th Septem-
ber 2013, Brno, Czech Republic, pp. 132−136, Brno.
Striebel, T., 2005: Höhlenbildung in ‘nicht
verkarstungsfähigen’ Gesteinen: welche Formen
sind Karstformen?- Laichinger Höhlenfreund, 40,
31–52.
Trimmel, H., 1968: Höhlenkunde.- Vieweg und Sohn, pp.
304, Leinen.
Vendeville B.C. & M.P.A., Jackson, 1992: e rise of dia-
pirs during thin skinned extension.- Marine and Pe-
troleum Geology, 9, 4, 331–353. DOI: 10.1016/0264-
8172(92)90048-J
Virlet, T., 1835: Observations faites en Franche-Comté,
sur les cavernes et la théorie de leur formation.- Bul-
letin de la Société géologique, 6, 154–164.
Weigand, A.M., Bücs, S.-L., Deleva, S., Lukić Bilela, L.,
Nyssen, P., Paragamian, K., Ssymank, A., Weigand,
H., Zakšek, V., Zagmajster, M., Balázs, G., Barjadze,
S., Bürger, K., Burn, W., Cailhol, D., Decrolière, A.,
Didonna, F., Doli, A., Drazina, T., Dreybrodt, J.,
ACTA CARSOLOGICA 51/1 – 2022
76

DEFINITION AND PROCESSBASED CLASSIFICATION OF CAVES
Ðud, L., Egri, C., Erhard, M., Finžgar, S., Fröhlich,
D., Gartrell, G., Gazaryan, S., Georges, M., Godeau,
J.-F., Grunewald, R., Gunn, J., Hajenga, J., Hofmann,
P., Knight, L.R.F.D., Köble, H., Kuharic, N., Lüthi,
C., Munteanu, C.M., Novak, R., Ozols, D., Petkovic,
M., Stoch, F., Vogel, B., Vukovic, I., Hall Weberg,
M., Zaenker, C., Zaenker, S., Feit, U., & J.-C., ies,
2022: Current cave monitoring practices, their va-
riation and recommendations for future improve-
ment in Europe: A synopsis from the 6 EuroSpeleo
Protection Symposium.- Research Ideas and Outco-
mes 8: e85859. DOI: 10.3897/rio.8.e85859
White, W.B., 1988: Geomorphology and hydrology of
karst terrains.- Oxford University Press, pp. 464,
New York.
White, W.B. & D.C. Culver, 2005: Denition of Cave.- In:
White W.B. & D.C. Culver (Eds.). Encyclopedia of
Caves. Elsevier, pp. 81–85, Amsterdam.
Woodward, H.P., 1961: A stream piracy theory of cave
formation.- Bulletin of the National Spelological So-
ciety, 23, 2, 39.
ACTA CARSOLOGICA 51/1 – 2022 77