When a Snail Dies in the Forest, how Long will the Shell Persist? Effect of Dissolution and Micro-bioerosion

2008·4.3 MB·English

by  Timothy A Pearce

Amer. Maine. Bull. 26: 111-117 (2008)
When a snail dies in the forest, how long will the shell persist? Effect of dissolution
and micro-bioerosion
TimothyA, Pearce
Carnegie Museum ofNatural History, 4400 Forbes Ave, Pittsburgh, Pennsylvania 15213, U.S.A., [email protected]
Abstract: Snail shells persist in the environment after death, but we know little about the rate at which shells decompose. Assumptions
aboutthe rateofshell decomposition are relevant to conservation biologistswho find emptyshellsorbiologists usingemptyshellsto make
mm
inferences about assemblages of living individuals. 1 put shells in 1.6 mesh litter bags (excluding macro-grazers) in Delaware and
northern Michigan, U.S.A. and monitored shell mass annually for 7 years. Decomposition rates differed among species, but 1 found no
difference in rates at two sites with different habitats. Surprisingly, loss ofperiostracum had no effect on shell decomposition rate. At the
locationsandhabitatsstudied,decomposition rateofsnailsaveraged6.4% peryear,excludingshellsthatbrokeduringtheexperiment (shell
halflife= 11.5years), or 10.2%, includingshellbreakage (halflife = 7.5 years). Halfliveswould likelybeshorter ifmacro-grazershad access
to shells. These results caution us to draw conclusions carefully when including empty shells in inferences about assemblages ol living
individuals.
Keywords: chemical weathering, death assemblage, decomposition rate, land snail, periostracum
After snails die, their shells persist in the environment. dead shells would overestimate the abundance of robust-
Although some shells survive as fossils for hundreds ofmil- shelled species. Using empty shells to calculate proportions
lions of years, most shells decompose (or effectively disap- of species in the assemblage of living individuals requires
pear) more quickly than that, probably on the order of assumingthat the death assemblageaccuratelyrepresents the
months or years. A shell in a dry, protected place such as a assemblage of living individuals. This assumption might be
desert or a museum might ptersist for hundreds o—fyears. A incorrect if different species, robust and fragile-shelled, de-
shell on the forest floor might persist more briefly but how compose at different rates or if shells at different sites de-
briefly? compose at different rates.
Although we know little about the decomposition rates Shells of Ovachlamys fiilgens (Glide, 1900) decomposed
ofland snail shells in leaflitter, many studies make assump- in an average of five months in Costa Rica during the dry
tionsabout therateofshell decomposition and could benefit season (Barrientos 2000). Aside from that study, most of
from more information about the decomposition rates (Me- what we know about the rate at which snail shells decom-
nez 2002). Management biologists making conservation de- pose is anecdotal. Welter-Schultes (2000) collected all the
cisionswould find decomposition rates relevant forknowing dead shells ofAlbinaria jaeckeli Wiese, 1989 that he could
how long ago a species was living at a site where an empty find in a particular area once in 1987 and again in 1990. The
shell was found. Although using data from snails collected number ofdead shells he collected was similar in each year,
alive would give more reliable results, biologists conducting suggesting that the dead shells had been completely replaced
biodiversity surveys commonly use empty shells as an expe- within three years.
dient way to indicate the presence of species at a site. Fur- Shells probably disappear by three main processes: dis-
thermore, since methods for recoveringsnails from leaflitter solution, breaking, and bioerosion (shell removal by graz-
{e.g., sieving and picking snails) are labor intensive, includ- ing). Shells that are protected from bioerosion decompose
ing information from empty shells is tempting for at least more slowly than shells that are exposed to this process
two reasons. First, empty shells can usually be recovered (Cadee 1999). Although consumption by larger organisms
alongwith live specimens with little extra effort, and second, such as other snails and decomposition by crushing and
the only occurrence of rare species in a sample might be breaking are real processes contributing to disappearance of
emptyshells, so excluding empty shells would discard infor- shells, in this study 1 excluded macro-grazers larger than 1.6
mm
mation. Studies in which empty shells and live shells are and breakage (for most analyses) by keeping target
counted indiscriminately would, of course, overestimate snails in mesh bags. Consequently, the shell decomposition
population sizes of the living snails. Furthermore, if shell rates in this studyare likelyto be slower than in experiments
decomposition rates differ among species, then including allowing access by macro-grazers such as other snails. How-
111
112 AMERICAN MALACOLOGICAL BULLETIN 26 • 1/2 • 2008
ever, excluding macro-grazers allowed me to focus on the the thin and fragile N. ovalis. I individually numbered shells
mm
effects of dissolution and micro-grazers less than 1.6 with India ink and divided specimens ofeach species evenly
that might remove shell material by grazing. into the 6 replicate bags (3 from each habitat), for example,
In this study, I address 4 questions: (1) Do shells of 8 A. alteniata in each bag. For species that did not divide
different species decompose at different rates, (2) Do shells evenly by 6 [e.g., N. ovalis and E. fraternum), I put one
in different habitat types decompose at different rates, (3) remainder in each of the outwash plain and the moraine
Does periostracum loss influence shell decomposition rate, localities. The Delaware bags each contained 7 Triodopsis
and (4) What is the mean half life ofa dead snail’s shell? fallax (Say, 1825).
At the start ofthe experiment, shells ranged from fresh
to eroded and some had been broken by small mammal
MATERIALS AND METHODS depredation. The fact that some shells were not fresh at the
start is not a problem because I compared relative shell loss
Localities and species fromyeartoyear. Although oldershellsmightbeexpectedto
To address how long empty shells persist in the forest, decompose more rapidly, for example due to periostracum
and to test for differences in decomposition rates among loss, as will be seen in the results, periostracum loss had no
species and among sites, I put individually numbered shells significant effect on percentage annual shell mass loss. The
in mesh litter bags at sites oftwo different habitats in north- Delaware shells were intact, but most were missing some
ern Michigan and at one site in Delaware with three repli- periostracum at the start of the experiment. In Delaware, I
cateslocated about 15 metersapart at each locality. The litter used only one species, Triodopsis fallax, which has a fairly
bags were approx. 22 x 22 cm with 1.6 mm screen openings robust shell.
and held about 1 literofsoil, leaflitter, andthedecomposing The fact that the mesh bags in Michigan and Delaware
shells. The litter bags were placed on the mineral soil surface had no species in common means that I cannot examine
and covered with about 5 cm of leaf litter and a small species-locality effects among all three localities. However,
amount ofsoil. I monitored shell mass annually for 7 years. since I used the same species at both Michigan sites, I can
Measuring mass loss from material in mesh bags has lookfor species-localityeffectsthere. Because sample sizes of
been used in manystudies ofleaflitter decomposition (Tay- some species were verysmall, caution should be exercised in
lor and Parkinson 1988) and is an applicable method to interpreting results.
studying snail shell decomposition. Choice of mesh size in
the bags is a trade-off between retaining fragments and al- Analyses
lowing entrancebygrazing organisms. Iflarger animalswere To determine shell decomposition, I weighed shells an-
important grazers on shells, then mesh sizes that exclude nually after retrieving them from the litter bags, cleaning off
them will result in slower shell decomposition rates. For adhering soil, and air-diying them to constant mass. Clean-
example, such a decrease in decomposition rate was ob- ing and drying resulted in little to no observable shell loss
served in studies ofleaf litter between larger mesh that ad- although a few small non-adhering pieces of periostracum
mitted and smallerthat excluded grazers (Cornelissen 1996). occasionally fell off shells during drying. I did not retrieve
The two northern Michigan localities included a mixed shell fragments less than about 4 mm^, so shells with these
pine and hardwood forest on a sandy outwash plain and a kindsofsmall fragment lossesare interpreted in thisstudyas
mixed hardwood forest on rich moraine soil. At these lo- shell mass loss through decomposition. Larger shell frag-
calities, I used locally collected shells from near the sites ments that could be associated with an individual shell were
where I studied their decomposition. The Michigan outwash included in the mass measurements of that individual. In
plain site tends to be drier because sandy soil does not hold addition, in order to examine mass loss by shell breakage
water as well. In Delaware, 1 used shells collected from the and to examine the effect of periostracum loss on decom-
Dehnarva Peninsula and put them in a beech-maple forest position rate, I also annuallyestimated the proportion ofthe
on piedmont. Soil pH measured 4.5 at all sites. shell and periostracum that were missing and measured
In Michigan, I used shells of7 species: Anguispira alter- maximum diameter ofthe remaining shell. I re-inked iden-
nata (Say, 1816) [n = 48), Discus catskillensis (Pilsbry, 1896) tification numbers onto the shells if the previous numbers
(n = 18), Euchemotremafraterinim (Say, 1824) (» = 8), Hap- had faded.
lotrema concavum (Say, 1821) {n - 18), Mesodou thyroidiis I made some adjustments to the data set. In some in-
(Say, 1816) (n = 18), Neohelix albolabris (Say, 1817) (ii = 6), stances, the apparent mass ofa shell increased over a previ-
and Novisuccineo ovalis (Say, 1817) (n = 2). I chose shells ous year (perhaps a piece ofsand had lodged in the shell).
that ranged in size from 4 to 25 mm diameter and ranged For instances in which the mass of a shell increased more
from the relatively robust and thick-shelled A. alteniata to than 10%, I removed the increased year from the analysis.
SHELL DECOMPOSITION 113
I excluded the first year ofDiscus catskilleusis measure- mass changed for individual shellsofA. alteruata, for 3 shells
ments from the analyses. The shells had been collected alive that remained intact, and 3 shells that experienced cata-
and driedwithout removingthe bodies. Theshells lost much strophic breakage at some point in the 7 year experiment, is
moremassthefirstyear (mean of5 mgor46% ofshell mass) shown (Fig. 2).
compared to subsequent years (0.5 mg or 15% of shell Shell decomposition rate differed among species in 119
mass), suggesting that the soft tissue body masses had been shell specimens that did not break (ANOVA, F= 3.774, P =
a significant part of the mass the first year. Indeed, Pearce 0.001 (Fig. 3). The Tukey post hoc test showed that decom-
)
and Gaertner (1996) reported dry mass ofD. catskilleusis to position rate ofintactAuguispira alteruata was less than that
beabout2 mg. Becausesoft part decomposition the firstyear ofD. catskilleusis, and M. thyroidiis was less than those ofD.
seemed to account for a large portion of the mass loss, I catskilleusis, H. coucavuui, and T. fallax. Interestingly, larger
excluded all first-year D. catskilleusis from mass loss analysis. shells had a slower percentage shell mass loss rate than
In most analyses, I was interested in the shell mass loss smaller shells (Pearson correlation, N= 142, - 0.088, P<
due to non-breakage factors rather than shell loss due to 0.001, unbroken shells only, not shown). Of the 5 species
breakage. To focus on effects of dissolution and micro- having at least 10 unbroken specimens, A. alteruata was the
grazers, I excluded from statistical analyses shells that suf- only one showing a significant within-species correlation of
fered catastrophic breakage during a particular year. I in- shell mass loss with shell size (Pearson correlation, N = 48,
cluded shells that were intact for at least 2 contiguous years. R~ = 0.099, P < 0.05), suggesting that it might be the major
I defined broken shells as those that lost more than 15% of contributor to the correlation for all species, although its
shell (estimated visually and recorded annually) or more
than 5% shell maximum dimension (measured and recorded WpahtetenrnI issubnjoetcticvoenltyracdliacsstiefdiedbyN.thaevatirlsenadnsdinH.otchoeurcasvpueuciiesa.s
annually). Defining shell breakage using the percent shell
relatively fragile shells and the rest as relatively robust shells,
present was independent of any changes in shell mass.
Although a control was not used in this experiment, for I saw no striking difference in trends for percent shell mass
loss.
example, to assess repeatability ofmeasurements from year
Shell decomposition rate did not differ significantly be-
to year, the precision of measurement obtained was much
tween the moraine site and the outwash plain site for 100
greater than the variation from year to year.
unbroken specimens in Michigan (ANOVA, F - 2.536, P =
Statistical tests and comparisons 0.114). Because different species decompose at different
The primary measure I used to assess shell decomposi- rates, and the shells in Delaware were different species from
tion was decrease in mass over time. In order to standardize those in Michigan, ifthere were differences between Michi-
so shells of different starting masses could be compared, I gan and Delaware specimens, I would not be able to differ-
calculated % shell mass loss over time and used this measure entiate species differences anci locality differences. Conse-
in comparisons. I used this measureforaddressingquestions quently, I omitted Delaware from the analysis comparing
comparing shell decomposition rates among different spe- shell decomposition rates among localities.
cies and different localities. A test for normality of percent Surprisingly, shells that lost more periostracum did not
shell mass loss ofAngiiispira alteruata showed the data to be decompose faster (Pearson correlation, R^ - 0.0004, P > 0.5)
kurtotic; sample sizes ofthe other species were too small to (Fig. 4). Although all values of mass loss per year greater
allow tests for normality. Consequently, I transformed the than 22% had less than 11% periostracum remaining, that
data using Log(x-l-l) and used ANOVA to compare different apparent greatervariability likely reflects the larger statistical
species or localities. I used the Tukey test to examine post- sample of shells with little or no periostracum remaining.
hoc differences. Furthermore, five species having sample sizes of at least 12
In order to evaluate whether shell decomposition rate individuals had shell decomposition rates independent of
increased after periostracum loss, I examined whether shell periostracum loss (separate species P-values > 0.2 to > 0.5).
mass loss rate correlated with percent periostracum loss. Although periostracum loss itselfdid not affect shell decom-
To calculate the half life ofthe shell, I extrapolated the position rate, it varied among species in 113 shells examined
shell decomposition rate to determine when half the mass (ANOVA, F = 4.997, P<0.0005) (Fig. 5).
would remain. Consideringthe intact shells only, which decomposed at
an average of 6.4% per year, the half life of an individual
RESULTS shell (protected from macro-grazers) would be 11.5 years
and after 35.8 years only 10% of the shell would remain.
Examples of shells that had decomposed for 4 and 7 Considering both intact and broken shells, which decom-
yearsareshown (Fig. 1). An exampledemonstratinghowthe posed at an average of 10.2% per year (Fig. 3), the half life
114 AMERICAN MALACOLOGICAL BULLETIN 26 • 1/2 • 2008
FNiog.ur0e31(.1A3.p6pemarmandcieasmeotfetrh;reAe,iBn)d,ivAi.duaalltesrhnealtlas aNfot.er142ye(a1r7s.6(lmefmt;coCl,umDn),,AM,esC,odEo)nanthdyr7oiydeiairssN(or.igh5t1c(o2l3u.1mnm,mB,, DE,,FF));.Anguispiraalternata
SHELL DECOMPOSITION 115
from dead snails has potential to bias estimates about as-
semblages ofliving snails. Admittedly, sample sizes of some
species in this study were very small, so results about those
species must be interpreted with caution; however, being
cautious with those results does not change conclusions
about species with larger sample sizes. Conclusions in stud-
ies using empty shells should indeed be drawn carefully.
Shell decomposition rates are likely influenced by a
plethora offactors. Three ofthe factors that might influence
shell decomposition rates are surface area to mass ratio, shell
robustness, and physical and chemical environment. Larger
shells, which likely have a smaller surface area to mass ratio,
lost mass more slowly than smaller shells in this study. Me-
Figure 2. Loss of Anguispira altcrnata shell mass over 7 years. nez (2002) also found that larger snail shells degraded more
Dashedlinesareshells thatbrokeduringtheexperiment;solidlines slowly. Such a size difference might be expected since shells
are shells that remained intact. with a high surface area to mass dissolve more rapidly
(Claassen 1998). Although physical and chemical destruc-
tion has been reported to be faster in thin-shelled than more
ofthe shell would be 7.5 years; after 22.4 years, only 10% of robust species (Evans 1972), no effect ofshell robustnesswas
the shell would remain. observed on shell decomposition rate in this study although
a better test for an effect of robustness needs to be con-
ducted. Robustnesswouldbe influencedbyshell thickness as
DISCUSSION well as form, such as ridges that add strength; future tests of
robustness should examine crush strength among species.
Because shells ofdifferent land snail species decompose Shell decomposition rates did not differ significantly at
at different rates, these results demonstrate that using shells two habitats in Michigan, despite the habitats differing in
substrate (sandy soil versus poorly
sorted moraine deposits), vegetation
intact + broken
bl (relatively low oak and pine with
20 - III _ sparser undergrowth versus taller as-
J48 18 “ pen forest with denser undergrowth),
i i I i and evidently moisture (although the
pH did not differ). This result con-
^ n0 trasts with leaf litter decomposition
rates, which are slower in sandy soil
w 20 having less moisture and less nutrient-
in intact
holding capacity (lohnson et al. 2000),
10 - 18 16 19 suggesting that processes regulating
42 i^ ^* ^i ^£ ^ ^ ^i leaf litter and shell decomposition
might differ. Because temperature,
1 1 / 1 1 1 1 \ 1 moisture, and pH likely play impor-
/
tant roles in shell decomposition,
/ /
shells in environments different from
c? 0 those I studied are likely to have dif-
O' species ferent decomposition rates. Indeed,
Barrientos (2000) found that shells in
Figure 3. Shell decomposition rate contrasted among species. Lower panel indicates decom- mesh bags (mesh size not stated) in
position rate for unbroken shells, upper panel the rate forboth broken and unbroken shells Costa Rica decomposed in an average
together. Numbers above bars indicate initial sample size. Bars with different letters within of 5 months.
a horizontal rowdiffered significantly; those without letters did not differ significantly. Data
Surprisingly, periostracum did not
frombroken shellswereincluded in thelowergraph onlyfortheirunbroken duration,which
seem to play a protective role in de-
explains unequal mass losses in top and bottom graphsforspecies havingsame sample sizes.
composing shells. Periostracum is usu-
116 AMERICAN MALACOLOGICAL BULLETIN 26 • 1/2 • 2008
more pervious after death, after having
stayed intact for years during the
snail’s life? Regardingthe second ques-
tion, the living snail might behavior-
ally or chemically maintain a good
bond between the periostracum and
the shell whereas the bond might
weaken after death, allowing ingress of
corrosive solutions.
Living snails would be important
grazers on decomposing snail shells.
Other micro- or meso-organisms that
Figure4. Shells that lost more periostracum did not decompose faster. Points showamount might graze on decomposing shells are
ofperiostracum remaining at the end ofa year (x-axis) and amount ofweight loss since the likely to exist. In a study ofshell mass
preceding year (y-axis). Individual shells can appear more than once for different years. loss ofHelix aspersa Muller, 1774 on a
dune area in the Netherlands, Cadee
(1999) found that shells protected
from bioerosion lost about 8% mass
in a year, similar to the rate found in
this study. However, in that study,
shells exposed to bioerosion by other
land snails lost mass much more rap-
idly, 34% in 70 days, indicating that
shells exposed to bioerosion could dis-
appear in less than one year. Ifmicro-
grazers are important contributors to
shell decomposition, then the soil
conditions [e.g.-, nutrients) would also
be relevant through their effect on the
micro-grazers.
Dissolution and chemical conver-
sion are often themaincontributors to
land snail shell decomposition (Claas-
Figure 5. Periostracum loss rate varied amongspecies. Numbers above bars indicate sample sen 1998). Colder water can dissolve
size. Bars with different letters within a horizontal rowofletters differed significantly; those more calcium carbonate (Claassen
without letters did not differ significantly. 1998) although more rapid dissolution
can be expected at higher tempera-
tures, at least in non-saturated water.
ally thought to protect shells from boring organisms, ero- On one hand, presence of moisture and lower pH increase
sion, or dissolution by leaflitter acids in terrestrial snails or the speed ofshell decomposition (Claassen 1998, Reitz and
acidic water in aquatic molluscs (Solem 1974). While most Wing 1999). On the other hand, alkaline soils rich in cal-
living shells have intact periostracum, the apices of some cium retard the breakdown ofempty shells (Claassen 1998,
living shells do erode over time. However, older molluscs Schilthuizen and Rutjes 2001, Cameron et al. 2003).
missing large areas of periostracum do not seem to suffer Although influences on decomposition rates of snail
serious erosion of the shell, suggesting that water and cor- shells on forest floors are poorly known, insights might be
rosion proofing qualities ofthe periostracum maybe onlyof gained from the more numerous studies on decomposition
secondary importance (Hunt and Oates 1978). Possiblyero- of leaf litter. Although different processes probably act on
sion soon after death starts on the inside surface ofthe shell, shells and leaves, results from studies finding leaflitter de-
which is not protected by periostracum. Nevertheless, two composition differences among habitats and climates are
questions remain: in the present study, why did the shell probably applicable to snail shell decomposition. For ex-
decomposition rate not increase after the loss of the peri- ample, warmer climates would probably increase decompo-
ostracum, and why does periostracum apparently become sition rates of shells, as it does in leaf litter (Bell 1974). In
.
SHELL DECOMPOSITION 117
leaves, factors most important at determining the rate of LITERATURE CITED
decomposition are those that regulate microorganism activ-
ity: temperature, moisture, nutrients, and energy source Barrientos, Z. 2000. Population dynamics and spatial distribution
(Berg and Ekbohm 1991). It microorganism activity plays a ofthe terrestrial snail Ovachlamys fidgens (Stylommatophora:
large role in shell decomposition, then shell decomposition Helicarionidae) in a tropical environment. Revista de Biologia
rates would also be largely affected by processes regulating Tropical 48: 71-87.
Bell, M. K. 1974. Decomposition of herbaceous litter. In: C. H.
microorganism activity.
The result that snail shells ofdifferent species have dif- pDoisciktiionns.oAncaanddemGi.cI.PrFe.ssP,ugLho,neddos.n.aBinodloNgyewofYPolrakn.tLPipt.teyriD-eBclom-
ferent decomposition rates has important ramifications for Berg, B. and G. Ekbohm. 1991. Litter mass-loss rates and decom-
studiesofendangeredspeciesandcommunityanalysis. Find- position patterns in some needle and leaf litter types. Long-
ing emptyshells ofan endangered species in habitats similar termdecomposition inaScotspineforest.VII. CanadianJour-
to those studied here would suggest that the species was nal ofBotany 69: 1449-1456.
living in the area within the last several decades at most. Cadee, C. G. 1999. Bioerosion of shells by terrestrial gastropods.
I However, the results ofthis study suggest that for commu- Lethaia 32: 253-260.
; nity analysis studies, using empty shells to infer abundances Cameron, R. A. D., M. Mylonas, K. Triantis, A. Parmakelis, and K.
j of the assemblage of living individuals might violate the Vardinoyannis. 2003. Land-snail diversity in a square kilo-
i metreofCretan maquis: Modest species richness, high density
assumption that the death assemblage accurately represents
and local homogeneity.JournalofMollnscan Studies69: 93-99.
I the assemblage of living individuals. Including empty shells Claassen, C. 1998. Shells. Cambridge Manuals in Archaeology.
I could overestimate the abundance of robust species, Cambridge University Press, Cambridge, U.K.
In the geographical locations and habitats I studied, and Cornelissen, ). H. C. 1996. An experimental comparison ofleafde-
j
with shells protected from macro-grazers, I extrapolate that composition rates in a wide range oftemperate plant species
shells will decompose to 10% of their former mass after and types. Journal ofEcology 84: 573-582.
i several decades. For practical purposes, e.g., in surveys re- Evans, J. G. 1972. LandSnailsinArchaeology. Seminar Press, London.
I covering shells from leaf litter samples, shells missing more Hunt, S. and K. Oates. 1978. Fine structure and molecular organi-
I than 50% of their former mass might not be findable or zation ofthe periostracum in a gastropod mollusc Buccinwn
i undatnm L. and its relation to similar structural protein sys-
identifiable, so shells in these conditions might effectively
I disappear in 7-12 years, their halflife. Halflives would likely tRoeymasliSnocoitehteyrofinLvoenrdteobnra(tBe,s.BiPohliolgoiscoaplhiSccailenTcreasn)sa2c8t3i:on4s17o-f45t9h.e
jI be shorter if macro-grazers had access to shells. Johnson, C. M., D. J. Zarin, and A. H. Johnson. 2000. Post-
Future studies might help tease apart the processes in- disturbanceabovegroundbiomassaccumulation in globalsec-
volved in shell decomposition. Exploring the decomposition ondary forests. Ecology 81: 1395-1401.
rates of shells in different environments (and geographic Menez,A.2002.Thedegradationotlandsnailshellsduringtheannual
localities) and noting biotic and abiotic influences would dry period in a Mediterranean climate. Iberus 20: 73-79.
help to address the importance ofdifferent situations (as has Pearce, T. A. and A. Gaertner. 1996. Optimal foraging and mucus-
been found in leaf litter decomposition studies) and of trail following in the carnivorous land snail Haplotrenia con-
cavum (Gastropoda: Pulmonata). MalacologicalReview29: 85-
scrapers or chemical weathering. Laboratory experiments
99.
could more directly evaluate the relative importance of the
three decomposition methods and the importance of pH Reitz, E. J. and E. S. Wing. 1999. Zooarchaeology. Cambridge Uni-
versity Press, Cambridge, U.K.
and temperature. Schilthuizen, M. and H. A. Rutjes. 2001. Land snail diversity in a
square kilometre of tropical rainforest in Sabah, Malaysian
Borneo. Journal ofMolluscan Studies 67: 417-423.
Solem, G. A. 1974. The Shell Makers, Introducing Mollusks. John
ACKNOWLEDGMENTS
Wiley and Sons, New York.
Taylor, B. R. and D. Parkinson. 1988. Aspen and pine leaf litter
decomposition in laboratoiy microcosms. II. Interactions of
I am grateful to AmyCortis, Jeffrey Firestone, and Erika temperature and moisture level. Canadian Journal of Botany
Martin forcollecting most oftheshells, to Alice Doolittle for 66: 1966-1973.
help with fieldwork, and to Joan McKearnon for advice Welter-Schultes, F. W. 2000. Approaching the genus Albinaria in
about the experimental design. Many thanks to Amanda E. Crete from an evolutionary point of view (Pulmonata:
Zimmerman for revising the figures. Collalaoration with Clausiliidae). Schriften zur Malakozoologie 16: 1-208.
Marvin C. Fields provided considerable help and encourage-
ment in analyzing and writing up the paper. Philip Myers Submitted: 23 December 2007; accepted: 20 |une 2008;
photographed Figs. lA, 1C, and IE. final revisions received: 4 September 2008