Role of Aerobic Microbial Populations in Cellulose Digestion by Desert Millipedes

Page created by Dolores Mendoza
 
CONTINUE READING
APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Aug. 1982, p. 000-000                                       Vol. 44, No. 2
0099-2240/82/080001-00$02.00/0

 Role of Aerobic Microbial Populations in Cellulose Digestion
                    by Desert Millipedes
                                            ELSA C. TAYLOR
             Department of Biology, University of New Mexico, Albuquerque, New Mexico 87131
                                 Received 15 January 1982/Accepted 22 April 1982

                                                                                                                     Downloaded from http://aem.asm.org/ on February 26, 2021 by guest
             I examined the role of aerobic microbial populations in cellulose digestion by
           two sympatric species of desert millipedes, Orthoporus ornatus and Comanchelus
           sp. High numbers of bacteria able to grow on media containing cellulose,
           carboxymethyl cellulose, or cellobiose as the substrate were found in the
           alimentary tracts of the millipedes. Enzyme assays indicated that most cellulose
           and hemicellulose degradation occurred in the midgut, whereas the hindgut was
           an important site for pectin degradation. Hemicellulase and 0-glucosidase in both
           species and possibly Cx-cellulase and pectinase in 0. ornatus were of possible
           microbial origin. Degradation of [14C]cellulose by millipedes whose gut floras
           were reduced by antibiotic treatment and starvation demonstrated a reduction in
           14Co2 release and 14C assimilation and an increase in 14C excretion over values
           for controls. It appears that the millipede-bacterium association is mutualistic and
           makes available to millipedes an otherwise mostly unutilizable substrate. Such an
           association may be an important pathway for decomposition in desert ecosys-
           tems.

   In desert ecosystems, rates of decomposition                 Cellulose decomposition is effected by three
are limited by available water, nitrogen, and                classes of enzymes: C1-enzymes (active upon
carbon (see literature cited in reference 20).               crystalline cellulose), Cx-erizymes (active upon
Decomposition and nutrient cycling are there-                noncrystalline cellulose and soluble derivatives
fore key processes affecting primary production              or degradation products of cellulose), and ,B-
in these arid regions. Cellulose decomposition, a            glucosidases or cellobiases (active upon cellu-
complex process mediated by a series of en-                  biose) (28). Because any of these enzymes may
zymes, is carried out by a wide variety of                   be produced by microflora or invertebrates, it is
organisms. In soils, decomposition can be ac-                important in quantifying cellulolytic activity to
complished directly through the activities of                trace the origin of these enzymes.
fungi (21) or aerobic and anaerobic bacteria                    The present study compares the levels of
capable of degrading cellulose to glucose and a              activity, origin, and ultimate function of cellulo-
mixture of acids (25). Indirect degradation is               lytic enzymes found in two sympatric species of
believed to be effected by the production of                 desert millipedes. Orthoporus ornatus (Spiro-
enzymes by microorganisms in invertebrate ani-               streptidae) is a large, desiccation-resistant milli-
mal alimentary tracts. Evidence of this is often             pede (13) that forages over a broad area. Coman-
conflicting and inconclusive, owing to difficul-             chelus sp. (Atopetholidae) is smaller, less
ties in culturing bacteria and distinguishing en-            desiccation-resistant, and more restricted in for-
zymes of microbial origin from those of inverte-             aging area and habitat. The digestive tract of 0.
brate origin. Nevertheless, some cellulases have             ornatus is composed of a small foregut (FG),
been fairly conclusively shown to originate in               larger midgut (MG), and very large, soil-filled
invertebrate animals (26; see references 23, 24,             hindgut (HG) which usually averages a little
and 38 for more conclusive evidence) and bacte-              over half the length of the animal (Fig. 1). In
ria (11, 30; see references 14, 16, 17, 36, 37, and          Comanchelus sp., the MG is large and soil filled,
41 for definite evidence). Studies of some milli-            whereas the HG is much smaller, although it is
pede species have indicated that cellulose is                also usually packed with soil.
digested during passage through the intestinal                  The following questions are posed in this
tract (7, 34); however, the origin of cellulolytic           paper. (i) Are bacteria in the guts of either or
enzymes is unknown. In another millipede spe-                both species capable of utilizing cellulose? (ii) If
cies, ingestion of a cellulose diet has been shown           cellulolytic enzymes are present in detectable
to result in midgut bacterial population develop-            amounts in the guts of these millipedes, in what
ment (2).                                                    part of the gut is such activity found? (iii) Is the
                                                       281
282     TAYLOR                                                                       APPL. ENVIRON. MICROBIOL.

                                                                                                                         Downloaded from http://aem.asm.org/ on February 26, 2021 by guest
  FIG. 1. Digestive tracts of 0. ornatus and Comanchelus sp. Top, Comanchelus sp. FG. 1.9 to 2.1 cm; MG,
2.1 to 4.2 cm; HG, 4.2 to 5.2 cm. Bottom, 0. ornatus FG, 0 to 0.9 cm; MG, 0.9 to 2.8 cm; HG, 2.8 to 6.8 cm.

activity probably of microbial origin, or is it            lose plates with 1% hexadecyltrimethyl ammonium
associated with intestinal tissue, indicating that         bromide (22). Zones of clearing around or under
these enzymes have a millipede origin? (iv) Do             colonies (which were scraped from the plates before
bacteria, through cellulose degradation, make              flooding) constituted a positive result, and those mor-
available to millipedes otherwise unutilizable             photypes which showed clearing were further charac-
                                                           terized morphologically (5). (Results are based upon
substrates, or are bacteria and millipedes com-            growth on CM-cellulose or cellulose medium, since
peting for the same food sources?                          some cultures would not grow on nutrient agar.)
                                                              The midsegment width (diameter of the middle of
            MATERIALS AND METHODS                          the millipede as measured with calipers) of each milli-
   Study site. Millipedes were collected from a site at    pede was measured, and then the FG, MG, and HG
the base of the volcanic escarpment immediately            were extracted under sterile conditions and triturated
northwest of Albuquerque, N.M. This Chihuahuan             separately in crucibles with 10 ml of sterile distilled
desert site is covered with large, dark, basalt boulders   water. Initial dilutions were shaken for 40 min on a
and a mixture of shrubs, mainly Gutierrezia sarothrae,     wrist action shaker (time predetermined from the
Atriplex canescens, and Rhus trilobata (see reference      standard curve of number of colonies versus time
35 for a more complete description of habitat). Before     shaken). Serial dilutions of each initial dilution were
each experiment, both species of millipedes were           spread on three replicate plates, and 1-ml aliquots
collected from the same habitat. Each species was          were added to five tubes for estimates of most proba-
then maintained separately in the laboratory on habitat    ble numbers. All inoculated media were incubated
soil and detritus food until used.                         aerobically at 30°C (mean daily field temperature dur-
  Bacterial isolation. Selective media were used to        ing feeding season) and 60% relative humidity for 14
isolate bacteria capable of producing any of the three     days (cellulose), 4 days (CM-cellulose), or 3 days
cellulases. Agar plates contained Skinner cellulose        (cellobiose). To facilitate counting, I air dried cellulose
medium B (33) with Hoagland trace element solution         plates for 2 h and then flooded them with dilute (1:10
and either 1% Whatman powdered cellulose with 1.5%         [vol/vol]) Safranin for 5 min, which yielded dark-
agar and cycloheximide (0.1 mg ml-') to inhibit fungi      orange colonies on a pale-orange background. Direct
(12) or 0.5% carboxymethyl cellulose (CM-cellulose)        microscopic counts were made in a Petroff-Hausser
with 1% agar (22). Whatman cellulose was purified by       counting chamber.
method of Leedle and Hespell (27). Because cellulose          To standardize data for millipedes of different sizes,
settled during solidification in the petri dishes, the     I measured the midsegment width of five individuals of
hardened agar disks were inverted before inoculation,      each species. The FG, MG, and HG were extracted
which gave bacteria a higher concentration of cellu-       and then dried at 60°C under vacuum for 48 h. A linear
lose. Medium for estimates of most probable numbers        regression of the midsegment width with the weight of
contained Skinner cellulose medium B, Hoagland             dried gut tissue plus contents gave standard curves
trace element solution, 1% cellobiose, and phenyl red      with high correlations.
indicator.                                                    Enzyme assays. To assess the presence in millipedes
   Medium containing Skinner cellulose medium B,           of cellulolytic enzymes and other enzymes capable of
Hoagland trace element solution, and 1.5% agar but no      degrading plant polymers, I maintained both species
cellulose, CM-cellulose, or cellobiose served as a         separately in the laboratory for 2 weeks before assay
control. In addition, I isolated numerically dominant      and gave them normal field detritus. Four 0. ornatus
morphotypes from cellulose and CM-cellulose plates,        and five Comanchelus sp. were pooled to make each
purified them by three subculturings, and then tested      sample. FGs of millipedes in each sample were re-
them for the presumptive ability to breakdown CM-          moved and triturated in a crucible with 0.001 M
cellulose by flooding 9-day old cultures on CM-cellu-      phosphate buffer, pH 7.0. For MG and HG samples,
VOL. 44, 1982                              CELLULOSE DIGESTION BY DESERT MILLIPEDES                            283
gut parts were slit open, and the contents were flushed       To determine the probable level of decrease in flora,
out with buffer and then triturated for assay of en-       I dissected the guts of four additional millipedes of
zymes. (Hereafter, MC will refer to MG contents and        each species that had received the starvation-antibiot-
HC to HG contents.) Tissue was flushed with two            ic treatment and determined bacterial levels by the
additional 10-ml aliquots of buffer (subsequently dis-     techniques described above.
carded) and then triturated in a 10-ml aliquot of fresh       Uniformly labeled, powdered ['4C]cellulose (specif-
buffer. (Hereafter, MT will refer to MG tissue and HT      ic activity, 7.7 ,uCi mg-'; dose, 1.4 mg per six milli-
to HG tissue.) All homogenates were centrifuged            pedes) was triturated with a glass rod in a small (75 by
(10,000 x g; 4°C; 20 min), and the supernatant fluid       12 mm) test tube; 0.15 ml of molten sterile agar was
was then passed through columns of PD-10 Sephadex          then added (potato dextrose for 0. ornatus and treha-
(5-cm bed height) with 0.001 M phosphate as the            lose-fructose for Comanchelus sp. [final concentra-
elution buffer. The protein-containing eluate fractions    tions: sugars, 0.5% each; agar, 1.5%]). Preliminary

                                                                                                                       Downloaded from http://aem.asm.org/ on February 26, 2021 by guest
were then pooled for use.                                  preference tests with various prepared media and
   Substrates were suspended in 0.1 M buffers that         mixtures of sugar and agar in distilled water indicated
reflected the mean pH of the gut part being assayed:       that 0. ornatus readily ingested potato dextrose agar,
for FG, phosphate buffer (pH 6.0); for MT and MC,          whereas Comanchelus sp. preferred a combination of
sodium acetate buffer (pH 5.5); and for HT and HC,         trehalose and fructose (unpublished data). Cellulose
Tris buffer (pH 8.5) (31). Assay procedures were those     was dispersed into the agar by mixing on a Vortex
of Martin and Martin (29). To determine the presence       mixer; drops were then quickly placed on tin foil with
of Cx-cellulase, hemicellulase, and pectinase, I made      a pipette. Each millipede (individuals were in separate
1% suspensions of CM-cellulose, locust bean gum,           Tupperware containers) was then offered a drop to eat.
and citrus pectin in appropriate buffers. A 0.3-ml         The cellulose could not be powdered finely enough to
portion of substrate was incubated with 0.3 ml of          be evenly dispersed throughout the agar, so the drops
enzyme for S to 60 min at 35°C. I terminated incuba-       (which also differed in size) did not contain a uniform
tion by adding 0.6 ml of 3,5-dinitrosalicylic acid rea-    quantity of cellulose. In addition, individual millipedes
gent (Bernfield reagent; 6) and heating the mixture in a   did not always ingest all of the agar drop offered. As a
boiling water bath for 5 min. A 0.9-ml portion of water    result of these problems, each millipede ingested a
was then added, and the optical density at 540 nm of       different and undetermined quantity of cellulose.
Bernfield reagent-reducing sugar complex was deter-           As soon as a millipede had eaten, it was placed in an
mined. Controls were run with enzyme denatured by          experimental flask in a train in which CO2 could be
heating in a boiling water bath for 15 min.                trapped. The flask in which a millipede with reduced
   I determined the presence of Cl-cellulase by incu-      flora was placed contained a moist soil-detritus layer
bating the enzyme with microcrystalline cellulose in       which had been autoclaved (20 lb/in2, 220°C, 1 h) and
buffer (50 mg ml-'). A drop of toluene, to inhibit         cooled before the millipede (surface sterilized by being
bacterial growth, was added before incubation for 24       dipped in 2% Lysol for 20 s) was added. This flask was
to 27 h with shaking at 35°C. Incubation was terminat-     preceded in the train by an air filter (15) to prevent
ed by rapid filtration through Celite; the assay condi-    bacteria from entering as the air was bubbled through
tions were those described above.                          the train and into the cocktail at the approximate rate
   A 3.32 mM solution of p-nitrophenyl-3-D-glucoside       of 30 ml s-1. "CO2 was trapped for 6-h periods in a
was used to determine the presence of aryl-f-glucosi-      cocktail containing 55% toluene, 39o ethylene glycol
dase. A 0.5-ml portion of the enzyme was incubated         monomethyl ether, 5.5% ethanolamine, and 0.5% PPO
with 0.5 ml of substrate for 5 to 60 min at 35°C.          (2,5-diphenyloxazole). Every 12 h, all fecal pellets and
Incubation was terminated by the addition of 1 ml of 1     5 ,ul of hemolymph were taken. Hemolymph was
M NH40H-NH4Cl buffer (pH 9.8), and the optical             mixed in a 99.5% toluene-0.5% PPO cocktail. Fecal
density at 420 nm of liberated nitrophenol was deter-      pellets were air dried, triturated, and suspended in a
mined. Controls were run with enzyme denatured as          Cab-o-sil cocktail (99.5% toluene, 0.5% PPO, 4% cab-
described above.                                           o-sil). The experiment was terminated after 54 h, when
   For the determination of the amount of protein          animals were dissected and two portions of fat body
present in each sample, the Bradford protein assay         per millipede were removed, weighed, solubilized in
(10) was used to run enzyme extracts and protein           NCS solubilizer, and mixed with a 99% toluene-0.5%
standards containing bovine albumin.                       PPO cocktail. Radioactivity was determined with a
   Radioisotope assay. Each species of millipede was       model LS230 liquid scintillation counter (Beckman
divided into two groups: insects with reduced flora and    Instruments, Inc.). Corrections were made for back-
control insects. Gut floras were reduced by a combina-     ground radiation levels and for counting efficiency and
tion of starvation and antibiotic treatment. Individuals   quenching among cocktails and among vials of each
that were newly emerged from dormancy in the soil          cocktail type (40).
and that had just started to eat were collected and           Control animals were treated in a similar manner
starved in the laboratory for 7 days (0. ornatus) or 5     except that starvation of these millipedes, which had
days (Comanchelus sp.). Preliminary tests indicated        been eating in the field for 2 weeks before collection,
that these lengths of time caused a gut flora reduction    was for 36 h. (Preliminary tests indicated that this
in each species without imposing undue physiological       length of time rendered millipedes hungry enough to
stress. During the last 36 h, individuals consumed         eat readily without affecting population levels of gut
drops of 10%o dextrose-water (0. ornatus) or 10%           flora.) Antibiotics were omitted from the sugar-water
fructose-water (Comanchelus sp.) (sugars were cho-         drops, millipedes were dipped in distilled water, and
 sen on the basis of species preference) with 0.044 mg     the experiment was run under unsterile conditions.
each of tetracycline and chlortetracycline per g of           Statistics. Bacterial and enzyme data for differences
 millipede weight.                                         between both species and within each species were
284     TAYLOR                                                                   APPL. ENVIRON. MICROBIOL.

                                                                                                                 Downloaded from http://aem.asm.org/ on February 26, 2021 by guest
              A                B                   A               B                 A               B
  FIG. 2. Number of bacteria from millipede guts cultured on three media. Values are for each milligram of gut
part tissue plus contents. A, 0. ornatus; B Comanchelus sp. Each vertical bar represents one standard error.
Results for 10 of each millipede species are shown.

analyzed by the Friedman nonparametric analysis of         nies on plates with cellulose sources, indicating
variance. Where differences were found, the Newman-        that most of the colonies on plates with cellulose
Kuels multiple range test was employed. Radioisotope       or CM-cellulose were probably using these car-
data and degree of flora reduction effected by antibiot-   bon sources. There was no significant difference
ic-starvation treatment were analyzed by the Wilcox-       between millipede species in overall numbers of
on rank sums test (46).
  Chemicals. All chemicals were purchased from Sig-        bacteria counted by direct microscopic counts
ma Chemical Co., with the following exceptions: the        or cultivated (Fig. 3A).
agar was from BBL Microbiology Systems; the potato            In pooling results for 0. ornatus organisms
dextrose agar was from Difco Laboratories, PD-10           isolated from all media, I found that the HG had
Sephadex columns were from Pharmacia Fine Chemi-           the highest number of bacteria per milligram,
cals, Inc., microcrystalline cellulose was from Poly-      followed by the MG and then the FG (all signifi-
science, [14C]cellulose and NCS solubilizer were from      cantly different at P < 0.001; Fig. 2). In pooling
Amersham-Searle, Cab-o-sil and ethanolamine were           data for gut parts and testing for density differ-
from Eastman Kodak Co., and toluene was from
Fisher Scientific Co.                                      ences among media, I found that cellobiose
                                                           supported larger numbers of bacteria (P < 0.05)
                                                           than did cellulose or CM-cellulose, which
                     RESULTS                               ranked equally. Each substrate supported the
   Bacteria. All data were standardized as de-             growth of the same number of FG bacteria
scribed above so that results could be expressed           (Table 1). Cellobiose supported significantly
as the number of bacteria per milligram of gut             higher numbers of MG and HG bacteria than did
tissue plus contents. Both millipede species con-          cellulose or CM-cellulose (probability:
VOL. 44, 1982

                _|°x-
                z
                  g
                 0;0
                        W.

                             42~
                             2--
                                    ~
                                     -3
                                          ~
                                           CELLULOSE DIGESTION BY DESERT MILLIPEDES

                                                ~     ~
                                                            -O R OO
                                                            U,RHOOU
                                                                      X
                                                           ~ ~ ~~&CMANCQHLLUh
                                                                                                8
                                                                                                               285

                                                                                                                      Downloaded from http://aem.asm.org/ on February 26, 2021 by guest
  FIG. 3. Direct microscopic counts of total numbers of bacteria from millipede guts. Values are for each
milligram of gut part tissue plus contents. A, Gounts of normal flora. Results for 10 of each millipede species are
shown. B, Gounts after gut flora reduction by starvation and antibiotics. Results for four of each millipede
species are shown. Each vertical bar represents one standard error.

highest numbers of bacteria in Comanchelus sp.            could be expected given its more restricted
grew in cellobiose media, followed by cellulose           foraging area, Comanchelus sp. had a far smaller
and then CM-cellulose media (all significant at P         diversity of bacteria than did 0. ornatus.
< 0.05). Equal numbers of FG bacteria grew in                Enzymes. Results of total enzyme activity
all three substrates (Table 1). The number of MG          assays indicated that both millipede species had
bacteria growing in cellobiose media was greater          the capacity to degrade cellulose and other plant
than the numbers growing in cellulose and CM-             polymers and that the MG was the main site for
cellulose media (P < 0.025; P < 0.001, respec-            cellulose and hemicellulose (but not necessarily
tively), whereas HG bacteria that grew in cello-          pectin) degradation in both millipede species
biose outnumbered those that grew in CM-                  (Table 3). The amount of pectin degradation in
cellulose only (P < 0.005).                               the 0. ornatus HG was surprisingly high. CG-
   Direct microscopic counts showed that for              cellulase activity was higher in 0. ornatus than
each millipede species (Fig. 3A), there was less          in Comanchelus (P < 0.0001); otherwise, total
bacteria in the FG than in the MG and HG,                 enzyme activity levels for the entire alimentary
which had equal numbers (P < 0.05).                       tract were the same for both species. Assay
   Many types of bacteria were able to establish          results for separated gut tissue and contents are
zones of clearing on CM-cellulose (Table 2).              expressed as micromoles of equivalent reducing
Many of these were, in addition, able to grow on          sugar liberated per microgram of protein in
media with cellulose, although no attempt was             enzyme homogenate per unit of time so that the
made to determine whether cellulose was uti-              results could be standardized with the least
lized, owing to difficulties in seeing clearing on        amount of bias: the millipedes used were of
such thick plates and to the drying of the plates         different sizes, and expressing results on a per-
during the long incubation period required. As            milligram-of-gut-weight basis (versus per micro-
                                                          gram of protein) would have strongly biased
  TABLE 1. Differences among bacteria from each           results toward the FG, MT, and HT, owing to
 gut part in ability to grow on three types of mediaa     their low weight, compared with the high weight
                                                          of soil present in the MC and HC (Table 4).
Gut p             Growth  of
                0. ornatus on:            Growth  of
                                    Comanchelus sp. on:
                                                            C1-cellulase in 0. ornatus was significantly
                                                          higher in MC than in HT (P < 0.001), FG (P <
  FG          CB, CEL, CMC            CEL, CMC, CB 0.05), or MT and HC (P < 0.05) (Table 5). (For
  MG          CB,CELCMC              CB, CEL, CMC clarity of discussion gut contents will be consid-
  HG          CB, CEL, CMC           CB, CEL, CMC ered as a separate gut part.) Activity of Cx-
   a Underlined variables are not significantly different cellulase in MC was higher than those in HT,
at P < 0.05. Relative values are given in the text. CB, MT, FG, and HC (probability:
286         TAYLOR                                                                APPL. ENVIRON. MICROBIOL.
                  TABLE 2. Morphology of bacteria able to clear media containing CM-cellulose
 Isolate             Colony morphology                                Cell morphology
      no.    Color  Configura- Marg b Eleva- Gram               Type and shape                Spore Motility
                      tion'     ri   tion' stain
       3d White        1        1      3     -     Streptococci in sheaths
       4d White        1        1      1     -     Large bacilli, round ends, single or chains of2 _-
       7d White        6        3      2     -     Long, thin bacilli, rounded ends, some curved    +
       8d White        6        4      7     -     Bacilli, single or chains of 2                   -
       gd White        1        3      2     -    Coccobacilli, single                                -   +
      13e White        2        2      7     -    Coccobacilli, irregular form                        -   +

                                                                                                               Downloaded from http://aem.asm.org/ on February 26, 2021 by guest
      14e White        2        2      5     -    Coccobacilli, single or clumped                     -   +
      16e White        1        1      5    +/- Coccobacilli, single                                  -   +
      25e White        1        1      3     _ Bacilli, short, clumped                                -
     2d Yellow         6        4      7     + Bacilli, single or branching                           -
     6d Yellow         5        2      6     - Bacilli, single or clumped                             -   +
   18d Yellow          1        4      7     + Cocci, single or short chains                          -
   21d Yellow          1        3      5     + Cocci, chains of .2                                    -
   23d Yellow          6        4      8     - Bacilli, long, thin, flexible                          -   +
    ise Yellow         1        1      2     - Bacilli, long, thin, curved, single                    -   +
    17e Yellow         1        1      3     + Coccobacilli, cornyiform, single or rows               -   +
     ld Pink           2        2      3     + Coccobacilli, single, rows, V-form                     -
   24d Orange          1        1      3     + Bacilli, V-form                                        -
   20d Orange          2        2      3     + Cocci, chains of -2                                    -
     5d Orange         1        1      3     - Coccobacilli, single or clumped                        -   +
    12 Orange          1        1      7     - Bacilli, single or clumped                             -
    19" Orange         1        1      3     + Coryniform, club-shaped rods in clumps                 -   +
    lld Beige          2        2      3     - Coccobacilli, single
  a 1, Round; 2, round with scalloped margin; 5, concentric; 6, irregular and spreading.
  b
     1, Smooth; 2, wavy; 3, lobate; 4, irregular.
  c 1, Flat; 2, raised; 3, convex; 5, umbonate; 6, hilly; 7, ingrowing into medium; 8, crateriform.
  d Isolated from 0. ornatus.
  e Isolated from Comanchelus sp.

MT all showed significantly higher activity than            dase and hemicellulase were present in extreme-
HT (probability:
VOL . 44, 1982                          CELLULOSE DIGESTION BY DESERT MILLIPEDES                           287

 TABLE 3. Total enzyme activity of each gut part        TABLE 4. Enzyme levels found in millipede gut
                                                                      tissue and contents
Enzyme and gut   Total enzyme activity (mean ± SE)'
      part           0. ornatus      Comanchelus sp.   Enzyme and gut       Enzyme activity (mean + SE)'
                                                            part           0. ornatus         Comanchelus, sp.
Cl-cellulase
  FG                1.18 ± 0.12        0.61 ± 0.09     C1-cellulase
  MG                7.33 ± 0.57        2.87 ± 0.76       FG              24.90    ±    8.87   142.99 + 112.96
  HG                1.68 ± 0.26        0.76 ± 0.10       MT              30.49    ±    2.49    27.25 + 5.78
                                                         MC              94.75    ±   11.56    78.26 + 28.32
Cx-ceHulase                                              HT               9.73    ±    2.74    61.59 ± 23.74
  FG               24.72 ± 3.16        3.98 ± 2.48       HC              34.57    ±    5.65    14.18 ± 8.72
                                      64.97 ± 10.12

                                                                                                                  Downloaded from http://aem.asm.org/ on February 26, 2021 by guest
  MG              275.01 ± 18.16
  HG               38.85 ± 3.84        3.44 ± 2.11     Cx-cellulase
                                                         FG              13.54 ±       3.92     8.23 t     8.23
1-Glucosidase                                            MT              10.37 ±       0.67     7.56 t     1.97
  FG               16.11 ± 1.81        4.69 ± 0.17       MC              69.14 ±       3.71    16.83 t     1.99
  MG              234.22 ± 14.72     271.39 ± 41.85      HT               1.64 ±       1.01     0.00 ±     0.00
  HG               28.87 ± 2.26        6.92 ± 1.33       HC              15.84 ±       2.88     0.00 ±     0.00
Hemicellulase                                          1-Glucosidase
  FG               47.89 ± 8.61       16.06 ± 1.87       FG                8.64   ± 1.41       10.73 t     1.63
  MG              317.18 ± 81.43     666.60 ± 47.70      MT                2.38   ± 0.29        3.85 t     0.54
  HG               35.19 ± 2.72        9.67 ± 0.52       MC               55.23   ± 8.03       88.84 ±     2.29
                                                         HT                2.17   ± 0.14        1.79 ±     1.11
Pectinase                                                HC               10.92   ± 0.68        8.36 ±     0.99
  FG                31.54 ± 9.59       18.82 ± 8.03
  MG               196.22 ± 27.75      66.88 ± 7.25    Hemicellulase
  HG               223.48 ± 52.01      42.75 ± 8.41      FG               28.32   ±    5.54    38.60   ±  9.56
                                                         MT                6.36   ±    0.61     9.59   ±  0.88
  a Values for C1-Cx-cellulases, hemicellulase, and      MC               78.56   ±    4.63   225.05   t 10.51
pectinase are expressed as micromoles of reducing        HT                3.19   ±    0.84     2.13   ± 2.13
sugar (x103) liberated per animal per minute. Values     HC               13.03   ±    2.58     6.64   ± 4.46
for [-glucosidase are expressed as micromoles of
nitrophenol (x103) liberated per animal per minute.    Pectinase
                                                         FG               17.53   ± 4.48     47.48 ± 20.11
                                                         MT                9.37   ± 1.39      9.36 ± 2.54
ments could be made [Table 7].) Impaired                 MC               46.62   ± 6.34     16.46 ± 1.20
cellulose degradation was shown by a decrease            HT               12.42   ± 3.93     12.57 + 3.70
in assimilation of label into hemolymph, a de-           HC               85.24   ± 44.34    54.03 ± 16.60
crease in production of 14Co2, and an increase in        a Values are expressed as follows: for Cl-cellulase,
the amount of label excreted. Levels of 4CO2           micromoles of maltose (x105) liberated per microgram
decreased over time, indicating that most of the       of protein per hour; for Cx-cellulase, hemicellulase,
label was either degraded or passed through the        and pectinase, micromoles of maltose (x105) liberated
gut during the course of the experiment. The           per microgram of protein per minute; for 13-glucosi-
percentage of total label assimilation in control      dase, micromoles of nitrophenol liberated per micro-
                                                       gram of protein per minute.
animals ranged from one-fifth to one-third, the
range of assimilation efficiency found by Woo-
ten and Crawford (44) for 0. ornatus (i.e., 22.8      Increase in bacterial growth on media with
± 2.65% for 0. ornatus and 26.58 + 3.33% for       cellulose, CM-cellulose, or cellobiose as the
Comanchelus sp.). Values for animals with re-      source of carbon and energy over growth on
duced flora were significantly (P < 0.05) lower:   control media indicates that the millipedes con-
16.25 ± 2.37% for 0. ornatus and 17.48 ± 3.24%     tained bacteria which could produce the Cl- and
for Comanchelus sp.                                Cx-cellulases and P-glucosidase necessary to
                                                   degrade cellulose. The bacteria in the FG were
                   DISCUSSION                      probably present as a result of ingestion and are
   Both millipede species are detritivores: they therefore not part of the resident flora. During
 feed on soil, plant litter, and other items found winter, the number of FG bacteria varies greatly
 on the soil surface. In addition, 0. ornatus from individual to individual, whereas the MG
 grazes on the bark of bushes in its habitat (45). and HG of all individuals retain large popula-
 These feeding patterns enable both species to tions (unpublished data).
 acquire varied microbial floras (fungal flora of     Numerical differences between bacteria cul-
 0. ornatus are discussed in another article tured and bacteria counted by direct microscop-
 [35a]).                                           ic counts indicate that many bacteria were not
288      TAYLOR                                                                     APPL. ENVIRON. MICROBIOL.
                        TABLE 5. Differences in enzyme production in each gut parta
                                                            Gut part producing enzyme
      Enzyme
                                                 0. ornatus                                    Comanchelus sp.
C1-cellulase                               MC HC MT FG HT                                   MC HT FG MT HC
CG-cellulase                               MC HC FG MT HT                                   MC MT FG HT HC
P-Glucosidase                              MC HC FG MT HT                                   MC FG HC MT HT
Pectinase                                  HC MC FG HT MT                                   HC FG MC HT MT
Hemicellulase                              MC FG HG MT HT                                   MC FG MT HC HT
  a
    Underlined variables are not significantly different at P < 0.05. Relative values are given in the text.

                                                                                                                    Downloaded from http://aem.asm.org/ on February 26, 2021 by guest
cultured by the methods employed. A large                     ornatus. However, although large numbers of
number of these may be strict anaerobes, which                bacteria were found in the Comanchelus HG,
have been found in the digestive tracts of other              the low levels of enzymatic activity, particularly
invertebrates (8, 39). In particular, the large,              CG-cellulase activity, indicate that this is not an
moist, soil-packed HG of 0. ornatus and the MG                important site of cellulose degradation. The abil-
and HG of Comanchelus sp. could quite possi-                  ity to degrade locust bean gum, which repre-
bly contain numerous anaerobic sites. Strict                  sents one category of hemicellulose (galacto-
anaerobes were not screened for in the present                mannan), demonstrates the capacity to degrade
study.                                                        some but not all categories of hemicellulose
  Both bacterial and enzyme data indicate that                which may be present in millipede food.
cellulose degradation occurred in the HG of 0.                   It is possible that hemicellulase and P-glucosi-
                                                              dase are of microbial origin in both millipede
                                                              species; CG-cellulase in the MG of 0. ornatus
 TABLE 6. Reduction in gut flora per milligram of             may also be of microbial origin. Both species
 gut tissue plus contents of antibiotic-treated versus        appear to be able to produce some pectinase
     control (36-h starvation) millipedes in various          which may, in addition, be produced by micro-
                     ecological groups                        organisms. Most cellulose and hemicellulose di-
   Ecological group and Gut % Reduction Significance          gestion occurs in the MG; pectin degradation
     millipede species      part              (P value)'      appears to occur primarily in the HG, which is
Cellulose degraders                                           generally considered (at least in the case of
  0. ornatus                FG      66.51        NS           insects) to function only in water and ion absorp-
                            MG      99.57       0.009         tion and to play no direct part in digestion (43).
                            HG      99.28       0.006         Recent work, however, indicates that bacterial
                                                              fermentation products produced in the HG of
  Comanchelus sp.        FG       97.18        0.02           cockroaches and the proctodeal dilation in scar-
                         MG       99.36        0.009          abid larvae can be absorbed from these sites into
                         HG       89.73         NS            the hemolymph (4, 9).
                                                                 Microorganisms present during incubation of
CM-cellulose degraders                                        the enzyme assay mixture may have contributed
  0. ornatus           FG         81.45         NS            some variance. This was particularly true for C1-
                       MG         99.86        0.006          cellulase because of the 24- to 27-h incubation.
                       HG         99.28        0.006          Microscopic examination of the assay mixture
                                                              revealed that although toluene inhibited microbi-
  Comanchelus sp.        FG       99.75        0.006          al growth, there were still large numbers of
                         MG       97.83        0.02           bacteria present. This may account for the ex-
                         HG       80.86        0.03           tremely high C1-cellulase activity in the Coman-
                                                              chelus FG.
Cellobiose degraders                                             It is difficult to establish beyond doubt wheth-
  0. ornatus             FG       97.77         NS            er these enzymes were of millipede or microbial
                         MG       99.92        0.006          origin. The fact that the activity in the gut
                         HG       99.34        0.006          contents was far higher than that in tissue may
                                                              point to a microbial origin. However, one cannot
  Comanchelus sp.       FG       99.64         NS             determine whether the low activity found in the
                        MG       99.94        0.006           tissue is due to millipede production or the
                        HG       99.08       0.06             incomplete flushing of gut contents from the
  aProbability based on Z statistic for Wilcoxon rank         tissue. In addition, since a secretory cell usually
sums test. NS, Not significant.                               produces and secretes its product constantly,
VOL. 44, 1982                            CELLULOSE DIGESTION BY DESERT MILLIPEDES                              289
             TABLE 7. Distribution of "4C in C02, hemolymph, fat body, and fecal pellet sampled
                                                14C distribution (mean % ± SE) in:
    Source                          0. ornatus                                      Comanchelus sp.
                       Control       Reduced flora        P value        Control        Reduced flora       P value
                       (n 6)            (n   5)'            vle          (n =6)            (n =5)0             au
CO2                 70.64 ± 5.89       47.06 ± 0.77        0.01      82.62 ± 2.77      58.89 ± 3.10         0.01
Hemolymphb           1.09 ± 0.27        0.51 ± 0.07        0.01        3.51 ± 0.70       1.75 ± 0.19        0.006
Fat bodyc            0.10 ± 0.02        0.09 ± 0.01        NSd         0.52 ± 0.19       0.26 ± 0.04         NS
Fecal pellete       27.36 ± 5.73       52.34 ± 0.75        0.01      12.51 ± 1.76      39.09 ± 4.16         0.006

                                                                                                                      Downloaded from http://aem.asm.org/ on February 26, 2021 by guest
  0 The gut flora of one individual had clearly not  been reduced so it was  excluded from  calculations.
  b Results are for a total of 20 ,u of hemolymph.
  c Results are means per milligram of fat body assayed.
  d NS, Not significant.
  I
    Means for five fecal pellets produced. Mean number of fecal pellets produced: 0. ornatus,               41.82;
Comanchelus sp., 80.75.

one would expect the level of activity to be          The fact that dependence on the Cl- and C,,-
lower in tissue than in contents. How great a cellulases is greater than dependence on ,-
differential would indicate a microbial rather glucosidase may be due to the nature of the food
than a millipede origin of enzymes is not known. ingested. The degree to which detritus is degrad-
For more conclusive determination of the enzy- ed by free-living microorganisms, particularly
matic origins, characteristics of enzymes from fungi, often determines the palatability of the
millipedes and cultured bacteria or levels of food to millipedes (19, 32). It may be, therefore,
activity in gut contents before and after treat- that much of the food that these millipedes
ment with antibiotics could be compared.            ingest has already been somewhat degraded by
   The importance of the microbial gut flora to free-living flora possessing the necessary Cl-
cellulose degradation is clearly suggested by the and C,-cellulases. This would also be a form of
radioisotope results. Increase in production of millipede dependence on microbially produced
"CO2 could be due to increased respiration by enzymes.
the millipedes or the bacteria. The increase in        The association between each species of milli-
label in the hemolymphs of control animals, pedes and its gut flora may be mutualistic.
however, does indicate that the products of Bacteria, through the production of cellulolytic
microbial degradation are assimilated by the enzymes, make available to millipedes other-
millipedes as well as by bacteria. Production of wise unutilizable substrates, which could be of
any of the three cellulases by bacteria could be crucial importance to millipede survival in des-
the important factor in allowing millipedes to erts where production of detritus is low (42). The
assimilate cellulose. The variability in body fat millipedes, in turn, provide for bacteria an envi-
results could have been due to the short duration ronment with regulated moisture, temperature,
of the experiment or to the fact that small, and pH and supply the bacteria with a constant
randomly selected portions were assayed, rather flow of substrates to degrade. This observation
than the total fat body. Comparison of the is supported by the fact that the gut contains far
percentage of assimilated label excreted by con- higher numbers of bacteria per gram than does
trol animals with the percentage excreted by the surrounding soil (unpublished data), indicat-
animals with reduced flora indicates that the ing that the gut provides a habitat in which
already low assimilation efficiency (although it is bacteria can grow and multiply. Further evi-
relatively high for a detritivore [44]) is further dence is provided by the finding that soil and
decreased by a reduction in gut flora.              litter bacteria increase in number during passage
   Comanchelus sp. and 0. ornatus are distantly through the digestive tract of the millipede spe-
related (different orders), yet both have evolved cies Glomeris marginata (1).
a strong dependence upon microorganisms for            In summary, the results indicate that both
digestive enzymes. This may be a universal millipede species contained bacteria capable of
phenomenon among millipedes, as other species utilizing cellulose. Enzymes necessary to de-
are also believed to rely upon microbial enzymes grade cellulose and other plant polymers were
(3, 34). Alternatively, this may be a function of present and may, in some instances, have been
living in a desert ecosystem where nutrients and of microbial origin. Reduction in gut flora result-
moisture are at low levels (18) and energetically ed in decreased assimilation of cellulose by
costly to obtain. Use of acquired enzymes would millipedes, indicating that products of microbial
decrease some of the expense of feeding for the cellulose degradation were utilized by milli-
millipede.                                                 pedes.
290       TAYLOR                                                                                 APPL. ENVIRON. MICROBIOL.

  This type of association has important impli-                          vol. 1, p. 291-292. Blackwell Scientific Publishers, Lon-
cations for nutrient cycling in deserts. Rates of                        don.
                                                                   13.   Crawford, C. S. 1972. Water relations in a desert milli-
decomposition are controlled by temperature,                             pede (Orthoporus ornatus, Girard) (Spirostreptidae).
moisture, and proximity of nutrients necessary                           Comp. Biochem. Physiol. 42A:521-535.
to sustain decomposer organisms (25). Condi-                       14.   Cruden, D. L., and A. J. Markovetz. 1979. Carboxy-
tions favorable to the activities of free-living soil                    methyl cellulose decomposition by intestinal bacteria of
                                                                         cockroaches. App. Environ. Microbiol. 38:369-372.
microorganisms may be sporadic and of short                        15.   Elsworth, R. 1969. Treatment of process air for deep
duration. Any association which would enhance                            culture. Methods Microbiol. 1:123-136.
the creation of a favorable decomposer situation                   16.   Evans, W. A. L., and E. G. Jones. 1962. Carbohydrases in
could greatly increase the rate of mineralization                        the alimentary tract of the slug Arion ater L. Comp.
                                                                         Biochem. Physiol. 5:149-160.

                                                                                                                                       Downloaded from http://aem.asm.org/ on February 26, 2021 by guest
and the production of humic compounds. Thus,                       17.   French, J. R. J. 1975. The role of termite hindgut bacteria
the presence of invertebrate-microbe associa-                            in wood decomposition. Mater. Org. 10:1-13.
tions could be of particular value during seasons                  18.   Friedman, E. I., and M. Galun. 1974. Desert algae, li-
when environmental conditions are inimical to                            chens and fungi, p. 166-212. In G. W. Brown, Jr. (ed.),
                                                                         Desert biology, vol. 2. Academic Press, Inc., New York.
the activities of free-living microbes and could                   19.   Gere, G. 1956. The examination of the feeding biology and
be of greater importance in more extreme des-                            the humicative function of Diplopoda and Isopoda. Acta
erts than in moderate deserts (35b). This funda-                         Biol. 6:257-271.
mental process would play a key role in making                     20.   Godall, D. W., and R. A. Perry (ed.). 1979. Arid-land
                                                                         ecosystems: structure, functioning and management, vol.
deserts habitable for higher trophic levels.                             1. IBP 16. Cambridge University Press, Cambridge, En-
                      ACKNOWLEDGMENTS                                    gland.
                                                                   21.   Gray, T. R. G., and S. T. Williams. 1971. Soil Micro-
   I thank E. Arguello and J. Washburn for technical assist-             organisms. Oliver and Boyd, Edinburgh.
ance. C. S. Crawford, M. M. Martin, D. E. Caldwell, J.             22.   Hankin, L., and S. L. Anagnostakis. 1977. Solid media
Trujillo, F. W. Taylor, W. W. Whitford, J. A. Wiens, and                 containing carboxymethyl cellulose to detect Cx-cellulase
G. V. Johnson are gratefully acknowledged for review of the              activity of microorganisms. J. Gen. Microbiol. 98:109-
manuscript.                                                              115.
  This research was supported by grants from the Graduate          23.   Hartenstein, R. R. 1964. Feeding, digestion, glycogen and
Research Allocations Committee and the Graduate School at                the environmental conditions of the digestive system of
the University of New Mexico.                                            Oniscus ascellus. J. Insect Physiol. 10:611-621.
                      LITERATURE CITED                             24.   Holden, M., and M. V. Tracey. 1950. A study of enzymes
                                                                         that can break down tobacco-leaf components. 2. Diges-
 1. Anderson, J. M., and D. E. Bignell. 1980. Bacteria in the            tive juice of Helix on defined substrates. Biochem. J.
     food, gut and faeces of the pill millipede, Glomeris mar-           47:407-414.
     ginata. Soil Biol. Biochem. 12:251-254.                       25.   Imshenetsky, A. A. 1968. Decomposition of cellulose in
 2. Baleux, B., and C. P. Vivarls. 1975. Etude preliminaire de           the soil, p. 234-255. In T. R. G. Gray and D. Parkinson
     la flora bacterienne intestinale de Schizophyllum sabulo-           (ed.), The Ecology of Soil Bacteria. University of Toronto
     sum var rubriper lat (Myriapoda, Diplopoda). Soc. Zool.             Press, Toronto.
     France 99:771-779.                                            26.   Lasker, R., and A. C. Giese. 1956. Cellulose digestion by
 3. Bano, K., D. J. Bagyaraj, and R. V. Krishnamoorthy.                  the silverfish Ctenolepisma lineata. J. Environ. Biol.
     1976. Feeding activity of millipede, Jonespeltis splendidus         33:542-553.
     Verhoeff and soil humification. Proc. Indian Acad. Sci.       27.   Leedle, J. A., and R. B. Hespell. 1980. Differential carbo-
     133:1-11.                                                           hydrate media and anaerobic replica plating techniques in
 4. Baylor, C., and J. Mathelln. 1980. Carbohydrate fermenta-            delineating carbohydrate-utilizing subgroups in rumen
     tion and by-product absorption studied with labelled cellu-         bacterial populations. Appl. Environ. Microbiol. 39:709-
     lose in Orycetes nasicornis larvae (Coleoptera: Scarabaei-          719.
     dae). J. Insect Physiol. 26:833-840.                          28.   Martin, M. M., and J. S. Martin. 1978. Cellulose diges-
 5. Benson, H. J. 1973. Microbiological applications, 2nd ed.,           tion in the midgut of the fungus-growing termite Macro-
     p. 145. William C. Brown Co. Publisher, Dubuque, Iowa.              termes natalensis: the role of acquired digestive enzymes.
 6. Bernfield, D. 1955. Amylases a and P. Methods Enzymol.               Science 199:1453-1455.
     1:149-150.                                                    29.   Martin, M. M., and J. S. Martin. 1978. The distribution
 7. Bocock, K. L. 1963. Digestion and assimilation of food by            and origins of the cellulolytic enzymes of the higher
     Glomeris, p. 85-91. In J. Doeksen and J. Van der Drift              termite, Macrotermes natalensis. Physiol. Zool. 52:11-21.
     (ed.), Soil organisms. North-Holland Publishing Co., Am-      30.   Nielsen, C. 0. 1962. Carbohydrases in soil and litter
     sterdam.                                                            invertebrates. Oikos 13:200-215.
 8. Bracke, J. W., D. L. Cruden, and A. J. Markovetz. 1978.        31.   Nunez, F. S., and C. S. Crawford. 1976. Digestive en-
     Effect of metronidazole on the intestinal microflora of the         zymes of the desert millipede Orthoporus ornatus (Girard)
     American cockroach, Periplaneta americana. Antimi-                  (Diplopoda: Spirostreptidae). Comp. Biochem. Physiol.
     crob. Agents Chemother. 13:115-120.                                 55A:141-145.
 9. Bracke, J. W., and A. J. Markovetz. 1980. Transport of         32. Sakwa, W. N. 1972. A consideration of the chemical basis
     bacterial end products from the colon of Periplaneta              for food preference in millipedes, p. 329-346. In G. N.
     americana. J. Insect Physiol. 26:85-89.                           Blower (ed.), Third International Congress of Myriapo-
10. Bradford, M. M. 1976. A rapid and sensitive method for             dology, Manchester. Academic Press, Inc., London.
     quantitation of microgram quantities of protein utilizing     33. Skinner, F. A. 1971. The isolation of soil Clostridia, p. 57-
     the principle of protein-dye binding. Anal. Biochem.              80. In D. A. Shapton and R. G. Board (ed.), Isolation of
     72:248-254.                                                       anaerobes. Society of Applied Bacteriology technical se-
11. Breznak, J. A. 1975. Symbiotic relationships between               ries 5. Academic Press, Inc., New York.
     termites and their intestinal microbiota, p. 559-580. In      34. Striganova, B. R. 1969. Cellulose decomposition in the
     D. H. Jennings and D. L. Lee, Sympiosis XXIX. Cam-                intestine of the millipede Pachyiulusfoetidissimus (Mur.)
    bridge University Press, Cambridge, England.                       (Juolidae, Diplopoda). Doklady Akad. Nauk SSSR
12. Buxton, A., and G. Fraser. 1977. Animal microbiology,              190:703-705. (In Russian.)
VOL. 44, 1982                                    CELLULOSE DIGESTION BY DESERT MILLIPEDES                                     291

35. Taylor, E. C. 1979. Seasonal distribution and abundance              Radiotracer methodology in the biological, environmental
    of fungi in two desert grassland communities. J. Arid                and physical sciences, p. 274. Prentice-Hall, Inc., Engle-
    Environ. 2:295-312.                                                  wood Cliffs., N.J.
35a.Taylor, E. C. 1982. Fungal preference by a desert milli-       41.   Wharton, D. R. A., M. L. Wharton, and J. E. Lola. 1965.
    pede Orthorporus ornatus (Spirostreptidae). Pedobiologia             Cellulase in the cockroach, with special reference to
    23:329-334.                                                          Periplaneta american (L.). J. Insect Physiol. 11:947-959.
35b.Taylor, E. C., and C. S. Crawford. 1982. Microbial gut         42.   Whitford, W. G., M. Bryant, G. Ettershank, J. Etter-
    symbionts and desert detritivores. Sci. Rev. Arid Zone               shank, and P. F. Santos. 1980. Surface litter breakdown in
    Res. 1:37-52.                                                        a Chihuahuan desert ecosystem. Pedobiologia 4:243-245.
36. Tetrault, P. A., and W. L. Weis. 1937. Cellulose decom-        43.   Wigglesworth, V. B. 1974. The principles of insect physi-
    position by a bacterial culture from the intestinal tract of         ology, 7th ed, p. 476-552. Halstead Press, New York.
    termites. J. Bacteriol. 33:95.
37. Thayer, D. W. 1978. Carboxymethyl cellulase produced           44.   Wooten, R. C., Jr., and C. S. Crawford. 1975. Food,

                                                                                                                                      Downloaded from http://aem.asm.org/ on February 26, 2021 by guest
    by facultative bacteria from the hind-gut of the termite             ingestion rates and assimilation in the desert millipede
    Reticulitermes hesperus. J. Gen. Microbiol. 106:13-18.               Orthoporus ornatus (Girard) (Diplopoda). Oecologia
38. Tracey, M. V. 1951. Cellulase and chitinase of earth-                20:231-236.
    worms. Nature (London) 167:776-777.                            45.   Wooten, R. C., Jr., C. S. Crawford, and W. A. Riddle.
39. Ulrich, R. G., D. A. Buthala, and M. J. Klug. 1981. Mi-              1975. Behavioral thermoregulation of Orthoporus ornatus
    crobiota associated with the gastrointestinal tract of the           (Diplopoda: Spirostreptidae) in three desert habitats.
    common house cricket, Acheta domestica. Appl. Envi-                  Zool. J. Linn. Soc. 57:59-74.
    ron. Microbiol. 41:246-254.                                    46.   Zar, J. H. 1974. Biostatistical analysis. Prentice-Hall,
40. Wang, C. H., D. L. Willis, and W. D. Loveland. 1975.                 Inc., Englewood Cliffs, N.J. 620 pp.
You can also read