Reproductive parameters of the Amazon river dolphin or boto, Inia geoffrensis (Cetacea:

Iniidae); an evolutionary outlier bucks no trends

Reproductive parameters of the Amazon river dolphin, or boto Inia geoffrensis, were estimated
from a population of individually recognizable animals in the Brazilian Amazon throughout 24
years. Gestation lasts 12.3–13 months, and calves are nursed for 1.5–5.8 years. The mean inter-
birth interval is 4.6 years, and there is no evidence of reproductive senescence. Females first give
birth at a mean age of 9.7 years and become sexually mature at body lengths of 180–200 cm. Body
length at birth averages 84 cm. The annual pregnancy rate was ~0.4, but the annual birth rate was
0.22; therefore, almost half of pregnancies do not result in a calf seen by our research team.
Entanglement of neonates in monofilament gillnets might account, at least in part, for these
losses. Births can occur year round, but peak at low water, when botos and their newborn calves
are concentrated on river margins. Despite profound physical and behavioural adaptations over
millions of years to a life in shallow, fresh waters and complex habitats, the boto has remarkably
similar reproductive characteristics to those of its marine counterparts, dolphins of the family
Delphinidae. The fundamental reproductive characteristics of small odontocetes have apparently
been robust to change over a very considerable evolutionary timespan.

Information on the reproductive parameters of most odontocetes is scant, to the extent that even
fundamental characteristics, such as reproductive rate and age at sexual maturity, are known
accurately for a minority of species. Historically, information was primarily gained from
examination of dead animals available from hunts, incidental catches in fishing gear or beached
carcasses (e.g. Harrison & Brownell, 1971; Perrin & Reilly, 1984; Martin & Rothery, 1993; Rothery
et al., 1995; Danil & Chivers, 2007) or from live animals held in captivity (Lacave et al., 2004).
However, in recent decades long-term studies of recognizable animals have yielded unique
information on reproduction in wild populations (Olesiuk, Bigg & Ellis, 1990; Wells & Scott, 1990;
Herzing, 1997; Mann et al., 2000; Gibson & Mann, 2008; Ward, Holmes & Blacomb, 2009;
Henderson et al., 2014; Wells et al., 2014; Arso Civil et al., 2017; Mann & Karniski, 2017). We
report here upon one such study of the Amazon river dolphin, or boto Inia geoffrensis (Blainville,
1817), in the western Brazilian Amazon. Commencing in 1994, this study has followed the lives of
657 individually identifiable botos (309 females) by routine observation and periodic physical
examination. This paper covers aspects of reproduction that are fundamental to understanding
population processes, including those needed to inform International Union for Conservation of
Nature (IUCN) Red List assessment. As such, it focuses on the reproductive history of females.
Hitherto, knowledge of reproduction in Inia has been based on opportunistic examination of
relatively small numbers of carcasses (Harrison & Brownell, 1971; Best & da Silva, 1984).

The ‘true’ river dolphins, of which the boto is one of only four species currently recognized
[although the Chinese river dolphin Lipotes vexillifer (Miller, 1918) may now be extinct], separated
from their marine counterparts in the Miocene (de Muizon, Lambert & Bianucci, 2017; Reeves &
Martin, 2017). Current evidence indicates that Inia may have diverged from the largest extant
toothed whale family, Delphinidae, the oceanic dolphins, as much as 18 Mya (Lambert et al.,
2017). Adaptation to shallow, warm, freshwater habitats with seasonal access to inundated forests
and lakes, no predators and a wide variety of readily accessible prey has resulted in profound
morphological, physiological and behavioural changes (Martin, da Silva & Rothery, 2008; da Silva &
Martin, 2017). Whether or not such changes extend to fundamental reproductive characteristics
has, in large part, remained unknown. This study provided the first opportunity to examine this
question systematically by virtue of access to a large number of river dolphins from one
population.

MATERIAL AND METHODS The work reported upon here was conducted in and within 50 km of
the Mamirauá Sustainable Development Reserve, Amazonas State, Brazil (3° S, 65° W; Fig. 1). The
reserve itself is a 11 137 km2 area of várzea habitat, comprising lakes, channels and seasonally
flooded forest. Most of the dolphins studied are encountered repeatedly over periods of many
years, but none is seen every month, and even the most frequently observed may apparently
leave the study area for several consecutive months. The population is consequently of unknown
size and ‘open’ to a greater extent than some study populations of small cetaceans elsewhere, e.g.
bottlenose dolphins Tursiops spp. in Florida, USA and New Zealand (Henderson et al., 2014; Wells,
2014) and killer whales Orcinus orca (Linnaeus, 1758) off the west coasts of Canada and the USA
(Olesiuk et al., 1990). The dolphins in the present study are a small part of a population that has a
continuous geographical distribution covering much of equatorial South America.

et al., 2014; Wells, 2014) and killer whales Orcinus orca (Linnaeus, 1758) off the west coasts of
Canada and the USA (Olesiuk et al., 1990). The dolphins in the present study are a small part of a
population that has a continuous geographical distribution covering much of equatorial South
America. The study consisted of two elements: short-term annual capture campaigns and year-
round daily sighting surveys. Each November, for 2–3 weeks, animals were trapped
opportunistically using nets in a narrow channel, and those selected for handling were then placed
on a stretcher and taken by fast boat to a nearby floating research facility. Here, they were held
for examination, marking and sampling for 2–15 min, then released. During this capture event (N =
1364 during 24 years), the sex of the dolphin was recorded, and it was weighed and measured.
Body length was measured in the standard way, between the tip of the rostrum and the tail notch,
with animals placed on a flat surface and the tape held parallel to the surface.

Two measurements were made during each capture event. If they differed by ≤ 5 cm (2 cm for
calves), the mean was recorded. If the measurements differed by more than this, further
measurements were made until two met the criteria, whereupon the outliers were disregarded.
The estimation of calf survival to 1 year of age required an estimate of the date of birth (DoB) of
calves in the sample. In most cases, the DoB was inferred from field observations of the mother
before and after the birth of the calf. In some cases, the body length of the calf was measured at
capture and used to estimate DoB by reference to a sample of measured calves whose DoB was
known to an accuracy of ± 2 weeks

Dolphins captured for the first time were given unique identification marks by freeze branding
(Wells, 2017). Very cold metal characters (usually two alphanumeric; 5 cm high for calves, 8 cm
high for juveniles and larger) were held against the skin on both sides of the dorsal surface for ~10
s. This treatment produced semi-permanent depigmented marks, which normally allowed
unequivocal identification in the field at ranges of ≤ 100 m. In addition, a small number of animals
with obvious scars or (human-caused) wounds were recognizable long term. Natural scars and
wounds were invariably superficial, healed quickly, and generally did not offer reliable
identification over multiple years

Most females were checked for lactation by applying light suction to the nipples using an open-
ended syringe. Pregnancy was detected by using mobile ultrasound equipment (SonoSite 180 PLUS
with transducer C60/5-2 MHz) to image the ovaries and uterus of potentially mature females, with
the dolphins lying on their side on a mattress. In most cases, pregnancy was confirmed by
measuring a visible fetus, but in a very small number of early pregnancies only a fluid-filled vesicle
was apparent (for details of the technique, see Lacave et al., 2004). Many study animals were
captured only once, but others were handled on ≤ 10 occasions. Capture and handling work was
licensed each year and monitored by the Brazilian Government agencies Instituto Brasileiro do
Meio Ambiente e dos Recursos Naturais Renováveis (IBAMA) and Sistema de Autorizaçao em
Biodiversidade (SISBio) (permit numbers 2344/9611- AC, 3552/9312-AC, 2002001.002344/96-11,
3462-1 to 13462-5, 13157-1 to 13157-4). The work was carried out in conformity with the Ethical
Committee of the Brazilian National Institute for Amazonian Research (INPA).

For the remainder of the year, field observations of marked and unmarked animals were made
from small boats on an almost daily basis, with the objective of collecting information on all
aspects of the lives of these animals visible at the surface. Between January 1994 and August
2017, > 66 000 records of identifiable dolphins were obtained during 5865 days of effort, each
including location, the identity of animals in the group and aspects of behaviour. Before 2010,
identifications were mostly made by eye, but from 2010 onwards almost all were confirmed by
photography. The frequency of field records of marked females varied substantially within and
between individuals, from none in a calendar year to as many as 115 (mean 12.3 year−1).
Additionally, post-mortem examination information was available from 45 dolphins found dead

Calves were ‘counted’ for the first time when an adult female was seen with an apparently new
calf on separate occasions (different days) or, less frequently, when the adult was seen repeatedly
with a tightly associated small calf during a period of observation of one dolphin group. It was
common for calves to appear at the surface beside an animal other than their mother, and
observers could not see behaviour below the surface because of water opacity, so the
requirement for independent sightings greatly reduced the risk of false-positive records of new
calves at the cost of inadvertently disregarding sole sightings of calves that subsequently died.

Analytical methods

The age at first parturition was determined for females captured and marked in their first few
years of life that were then observed frequently and went on to give birth to a calf. It is possible
that some first births in this sample of 22 adults were missed, so the value given must be
considered to be a maximum. The age of calves at first capture was estimated from field
observations of the mother and/or by reference to a growth curve constructed from length-at-
capture data of calves whose date of birth was known accurately from observational data. The
birth date of observed calves was assumed to be the mid-date of the period between their mother
being last seen conclusively without the calf and first seen with the calf. Calves were included in
the sample to determine seasonality of birth only when this period was a month or less, so the
error could be no more than 15.5 days.

Inter-birth interval

In many cases, the date of birth of a recorded calf was not known with any precision and, because
births could occur year round, even the year/season of birth was not always known for calves that
were not captured with their mother and measured and marked. Consequently, the interval
between two consecutive births of a mother was rarely known directly, although the presence of
those calves at some time during the nursing period would probably be apparent in the case of
mothers that were seen frequently, especially if the calf survived more than a few months after
birth. The mean collective inter-birth interval (IBI) was therefore calculated as the reciprocal of the
collective calving rate, adjusted for the fact that the beginning and end of the period during which
a female was monitored for new calves (the monitoring period) could occur at any time in the
reproductive cycle

Multiple simulations of sampling a plausible range of reproductive histories demonstrated that the
unadjusted calving rate (number of calves seen/time elapsed) for an adult female, calculated over
a monitoring period starting and finishing at random, was positively biased. This was simply
because all calves seen during the monitoring period were ‘counted’ but, on average, the
monitoring period commenced halfway through the dependent period of the first calf and ended
halfway through that of the last calf in the series. The bias was removed, on average, if a period
equating to the IBI was added to the monitoring period for each animal. In the case of females
that were monitored from a young age and were known to be reproductively mature only when
seen with their own calf (usually in its first few months of life), the monitoring period commenced
at that time and finished randomly (e.g. at their disappearance or the end of the study period). In
this case, the appropriate adjustment was to add 50% of the mean IBI for this class of female to
the monitoring period for each animal in the sample. Thus, for mature females:

The above equations were solved iteratively, with each iteration using the IBI estimate from its
predecessor. Both quickly converged when initiated with an estimate of the appropriate IBI, and
the results were insensitive to the initial estimate.

RESULTS

Age and length at sexual maturity

The mean age at first birth was 9.7 years (SE = 0.33, N = 22, range: 7.2–12.4). Given that gestation
lasts ~1 year (see below), the mean age at first ovulation is no more than 8.7 years, but may well
be less than this (because some ovulations may not result in pregnancy). The shortest females
found to be pregnant or lactating when captured were 180 cm long (N = 3). Twentynine females
captured at lengths between 180 and 201 cm inclusive were deemed to be non-reproductive on
the basis of physical examination, appearance and, in many cases, field observation (e.g. usually
seen in proximity to their mother). Although some of these animals could plausibly have been in
the very early stages of pregnancy at capture, it seems reasonable to conclude that females first
become pregnant at lengths between ~180 and 200 cm. The mean asymptotic body length of
females in this population is 199.8 ± 1.22 cm (range 182–218 cm) (Martin & da Silva, 2006);
therefore, first pregnancy occurs at between 90 and 100% of this length.

Litter size

In the ultrasound examinations where the number of fetuses could be imaged, a single fetus was
detected in 81, and multiple fetuses in none. In no case among > 300 was more than one
dependent calf seen with a mother. The boto is monotocous/uniparous.

Duration of gestation

A lower bound on this measure was provided by the fact that no pregnancy detected by
ultrasound examination resulted in a recorded calf in < 9.9 months, but three calves were seen
between 9.9 and 10.6 months after pregnancy was detected. The size of the first of these fetuses
(thoracic diameter of 18.5 mm on 13 November 2015 in adult female no. 417) would, if it had
been that of a similar-sized odontocete with a 12.2 month gestation period (Tursiops), indicate
that it had been conceived 90 days earlier (Lacave et al., 2004). This calculation, in association with
the field observations of the resulting calf (first seen on 8 September 2016 and frequently
thereafter, but it could have been born up to 14 days earlier), predicts a gestation period for this
fetus of 376–390 (383 ± 7) days or 12.6 ± 0.3 months. Given that this range is a little above the
mean gestation period of Tursiops, the time since conception of this fetus might have been slightly
under-estimated, but it seems likely that the birth of this calf occurred between 12.3 and 13
months after conception. An upper bound for the duration of gestation was provided by the fact
that none of the females examined in consecutive years (in November) was pregnant on both
occasions. Had any been so, the fact that most pregnancies were detected late in the first
trimester or in the second trimester would have implied a gestation of ≥ 14 months. An
independent upper bound was provided by the minimal period between a female losing a young
calf prematurely and being seen with a new calf, which was 15.0 months. This period necessarily
included time for the mother to come into oestrus again and was also extended by the time
between the first calf being last seen and dying and between the second calf being born and first
recorded. These bounds offer little precision but are at least consistent with the 12.3–13 month
estimate for the calf of female no. 417, and this single calf provides the most precise estimate for
the duration of gestation in Inia.

Size at birth

The botos of shortest body length at capture measured 87, 89 and 93 cm, and in no case had the
mother been seen with the calf previously. The shortest animals found dead with evidence that
they had been born alive (e.g. gut contents, net entanglement marks) were 78, 85 and 94.5 cm in
length. The longest fetuses in pregnant females collected for post-mortem examination after being
reported dead were 84 and 87.5 cm in length. On this basis, we conclude that most births occur at
body lengths of 78–90 cm (mean 84 cm). At this length, calves weigh ~12 kg, which is 12% of the
mass of a non-pregnant female of adult asymptotic length (Martin & da Silva, 2006).

Pregnancy rate

In 205 examinations of sexually mature females, pregnancy was reliably detected in 86 (42.0 ±
6.8%). Of the 85 animals in this pregnant sample checked for signs of lactation, 30 (35.3%) were
still producing milk for the previous calf. Most pregnancies were estimated to be late in the first
trimester or in the second trimester. Fifty-four females (26.3 ± 6.0%) were neither pregnant nor
lactating. Assuming that gestation lasts 12.3–13 months (see ‘Duration of gestation’ above), the
detection of an established pregnancy in 42.0% of examined mature females implies that between
38.8 and 41.0% of mature females become pregnant each year (annual pregnancy rate).

Duration of lactation

Of the lactation periods leading to a surviving calf (i.e. the lactation was not prematurely truncated
by the death of a calf) known accurately, the minimum was 1.5 years and the longest 5.8 years
(mean 2.8 years, SE = 0.29, N = 87). Seven lactations were known to last 50 months (4.2 years) or
more.

Maximal age

The female of greatest minimal age in this study was at least 32.3 years old when sighted in
November 2017, assuming that she was as young as the earliest known calving female (7.2 years)
when producing her first known calf, which had an estimated age of 3.0 years in November 1995.
Other animals in this study, also captured for the first time as mature adults, might have been of
greater age.

Seasonality of reproduction

Figure 2 shows the timing of estimated birth dates, which are normally distributed (Shapiro test; P
= 0.64 N = 53). A clear peak is evident, such that 53% of births are in the 90 day period between 23
July and 21 October, but births can occur throughout the year. The peak date is 6 September.

Inter-birth interval and annual birth rate

The mean IBI was 4.56 years (N = 71 reproductive females, excluding animals that never produced
a calf to our knowledge). This value is derived from the aggregate of 21 animals that were first
marked when immature, with an average IBI of 3.62 years, and 50 animals first marked as mature
adults, with an average IBI of 4.96 years. The monitoring period for each animal (i.e. the period of
study during which sightings of the female were sufficiently frequent to allow a high probability of
detecting a calf) was no less than 5 years (maximum 23.6 years, mean 11.9 years). The reciprocal
of the IBI is the proportion of mature, reproductively active females that give birth in a year, on
average. This is termed the annual birth rate, and it has a value of 0.219. It is important to note
that this IBI is based on calves that survived long enough to be seen by our research team; usually
to between 1 and 11 months from birth. Calves that died as neonates would, in most cases, not be
recorded. Consequently, the IBI as calculated is greater than it would be if all calves were known
from birth. Likewise, the annual birth rate is negatively biased.

The minimal period between births of consecutive calves, when the first was successfully weaned,
was 2.0 years. The longest period between known births was 7.7 years, but it is possible that
another calf was born during this period and died before being seen

Survival to 1 year of age

Of 226 calves that were first noted (or captured) at < 1 year of age, 180 (79.6 ± 5.2%) survived to 1
year of age. It is important to note that this is not a measure of survival from birth to 1 year of age,
but from discovery of the calf to 1 year of age.

Sex ratio at birth

Very few dolphins were captured within days or weeks of birth, so an estimate of this measure
was sought from within the pool of the smallest animals captured during the study, all known to
be < 1 year of age. Figure 3 shows how the apparent sex ratio varies as the sample of animals is
increased by adding others in order of increasing body length. The graph demonstrates how the
sex ratio varies widely at small sample sizes, as would be expected. However, contrary to
expectations the sex ratio does not stabilize as the sample size increases; it moves from 0.34 to a
peak of 0.53 in an apparently directed manner as the sample increases from 35 to 87, before
diminishing again. In these circumstances, the natal sex ratio cannot be inferred robustly from
postnatal dolphins.

Reproductive senescence

Of 21 females marked during or before 1998 and still alive in September 2017, 19 (90%) had been
actively reproducing (pregnant and/or lactating) within the previous 3 years (since September
2014). All these females were known to be ≥ 20 years old in 2017, and many were > 25 years of
age. Only three of 76 mature females known to have been alive over the decade ending in August
2017 did not produce a calf within that period to our knowledge. The potential maximal lifespan of
females is not known, not least because many die as a result of interactions with fisheries rather
than from natural causes, but on current evidence there is no indication of reproductive
senescence in this species.

Figure. 3. Neonatal sex ratio (percentage of males), as predicted by the smallest botos measured
at first capture. The x-axis shows the number of animals included in the sample. The red line
represents sexual parity. The sample starts with the smallest boto ever measured (N = 1). The
second-smallest animal is then added (N = 2), then the third-most (N = 3) and so on. The ratio is
erratic at sample sizes < 20, reaches a minimum (0.34) at N = 35 and a maximum (0.53) at N = 87.
However, the ratio never truly stabilizes.

many were > 25 years of age. Only three of 76 mature females known to have been alive over the
decade ending in August 2017 did not produce a calf within that period to our knowledge. The
potential maximal lifespan of females is not known, not least because many die as a result of
interactions with fisheries rather than from natural causes, but on current evidence there is no
indication of reproductive senescence in this species.

DISCUSSION
The results given in the present paper, gained from thousands of hours of observation of > 650
dolphins during a period of 24 years, provide the first opportunity to understand the reproductive
characteristics of the Amazon river dolphin and to compare this evolutionary outlier with other
cetaceans. Although based on the best available data at the time, previous estimates of Inia vital
rates can now be seen to be substantially inaccurate in some important respects (e.g. Taylor et al.,
2007). It is important to recognize, however, that long-term studies of live cetaceans, even when
they include periodic physical examinations, as here, cannot provide detailed knowledge of every
aspect of reproduction. Furthermore, even a study period of 24 years is too short to allow the
examination of some aspects of biology in a species that can clearly live much longer than this.

Determining the duration of gestation in cetaceans is extremely challenging, and it has been
accurately established only in the case of captive animals, where mating opportunities can be
controlled and births observed. Indeed, for only two genera has such information been published;
Tursiops (Lacave et al., 2004) and Orcinus (Duffield et al., 1995). In most other cases, the period of
gestation has been estimated from fetuses examined at post-mortem examination based on
certain assumptions (e.g. Danil & Chivers, 2007) or, as in the present study, is inferred from inexact
field data. The difficulties are exemplified by the fact that despite decades of daily observations,
the carrying out of many ultrasound examinations of pregnant dolphins and the recording of > 300
different calves in the field, in only three cases can the period of gestation in botos be narrowed
down to within a few weeks, and then only by assuming that the growth rates of Tursiops and Inia
fetuses are very similar. Nevertheless, the evidence for a gestation period of some 12.3–13
months in this species is compelling, and it is striking only in how similar it is to estimates of the
same measure in dolphins of a similar size but very different type, namely the marine family
Delphinidae.

The relatively low degree of birth synchrony found in the boto, which lives year round at low
latitudes, where temperature and daylight length are fairly constant year round, is of no surprise
(cf. ungulates: English et al., 2012). The implication is that many mature females ready to ovulate
do not wait until the next seasonal peak before doing so and becoming pregnant. Given the high
level of losses of neonates or fetuses discovered in this population, this flexibility could be
important. Indeed, these losses could well be driving the lack of breeding synchrony observed. The
implication for adult males is clear; that mating opportunities might occur at any time of year.
Albeit with small sample sizes, neither Best & da Silva (1984) nor da Silva (1994) detected seasonal
variation in the testes of adult Inia.

Although seasonality of reproduction is more pronounced in some oceanic dolphins than in botos,
even this characteristic seems not to differentiate the two families; delphinid populations living in
the eastern tropical Pacific [the short-beaked common dolphin Delphinus delphis (Gray, 1828),
pantropical spotted dolphin Stenella attenuata (Gray, 1846) and spinner dolphin Stenella
longirostris (Gray, 1828)] have no more birth seasonality than the boto (Barlow, 1984; Danil &
Chivers, 2007).
The peak of boto calving coincides with the period of lowest water levels in the Amazon basin,
when almost all animals are concentrated along the margins of the main rivers. Within weeks,
however, rising waters allow dolphins to enter the lakes and channels of the flooded forest.
Lactating mothers with their young calves are quick to take advantage of these areas of low
current and plentiful small fish (Martin & da Silva, 2004); conditions which are very likely to benefit
vulnerable small calves as they learn to hunt. With gestation slightly more than a year, most
conceptions also occur during the period of dolphin aggregation, coinciding with a period when
socio-sexual object-carrying displays are at their peak (Martin et al., 2008).

The discovery that ~40% of mature females become pregnant each year contrasts with the mean
IBI of 4.56 years. If every pregnancy resulted in a calf recorded by our observers, the expected
annual pregnancy rate would be close to the reciprocal of the IBI, or 0.219 (21.9%), i.e.
approximately half of what was observed. The implication is that only a little more than half of the
pregnancies extant in November (when the dolphins are examined) produce a calf old enough to
be noted by our field team. Furthermore, given that 80% of observed calves reach 1 year of age,
only 43–46% of November pregnancies produce a 1-year-old calf. Wells et al. (2014) illustrated
equivalent values for fetal and neonatal survival in Sarasota Bay (FL, USA) bottlenose dolphins. Ten
of 12 pregnancies (83%) resulted in a known calf, and 67% of detected pregnancies survived to 1
year of age. In comparison, our study population clearly suffers very high mortality of fetuses in
the second half of gestation and/or of neonates in the first weeks or months of life. The prospects
for population sustainability or growth are clearly very different between these two studies.

An obvious question is whether the capture and handling of the mothers could, in itself, cause
fetuses to be aborted and result in this high loss rate. The evidence, however, shows that this is
not the case. Firstly, capture in any year has no impact on the reproductive output of females in
this study (Martin, da Silva & Rothery, 2006). Secondly, the annual birth rate is calculated from
field observations of all recognizable females, of which only a small proportion are handled in a
year and many only once in their lifetime.

Perinatal mortality is often high in cetaceans (Whitehead & Mann, 2000), and some would be
expected in utero and at birth in any wild mammal, but the scale of the losses at this stage of
development in botos is striking. Similar poor fetal/neonatal survival has been demonstrated in
only one other wild dolphin population, and the cause was almost certainly local human impact.
Pregnancy success in this population of bottlenose dolphins was one-quarter that of another on
the east coast of the USA that was not subject to the same impact (Lane et al., 2015). Available
evidence suggests that human activity might also be implicated in reproductive failure in botos.
Although the species has no natural predators, the humandominated environment into which
neonates are born today is dangerous. Monofilament gillnets are commonly deployed in the
Amazon, at densities of up to 5 km−1 of river bank in the study area (authors’ personal
observations), and young dolphins are neither experienced enough to avoid them nor strong
enough to escape once enmeshed. Net-killed or harpooned young botos are occasionally found by,
or reported to, our field team, but only a small proportion of carcasses are likely to be reported or
encountered. Most are used by fishermen as bait for the catfish Calophysus macropterus
(Lichtenstein, 1819) (Loch, Marmontel & Simoes-Lopes, 2009; da Silva, Martin & do Carmo, 2011;
Iriarte & Marmontel, 2013; Brum et al., 2015) or quickly consumed by scavengers, especially
caimans Melanosuchus niger (Spix, 1825) and Caiman crocodilus (Linneaus, 1758), and piranhas
(Serrasalmidae).

The fact that 80% of calves first recorded by our field team at < 1 year of age survived to 1 year
appears to contradict the suggestion that neonatal mortality may be high. However, calves are
especially vulnerable when very young (e.g. Henderson et al., 2014), before most are first
encountered by researchers, so the proposition is still reasonable. In this study, calves are not
‘counted’ until their mother is unequivocally seen in close association with a small calf on at least
two occasions. Although this approach avoids false positives attributable to temporary
associations with the calf of another female, it must also result in false negatives, i.e. true calves
being ignored on the one occasion that they are witnessed before dying. Female dolphins that
were known to be pregnant in November of the previous year are often photographed once with a
small calf at around the time that the birth would be expected, but thereafter without a calf.
Sometimes the calf is known to be that of another female, but usually it is unmarked and could
belong to any female in the group, including the one we are expecting to give birth.

On one occasion, a female was encountered with a freshly dead neonate, which she was
attempting to support at the water surface. The probability of observers being in the right place at
the right time to witness that event was very small; this chance encounter demonstrated that the
birth of neonates that survive for only days or weeks might well pass unnoticed.

Over a third of pregnancies in this species commence while the mother is still producing milk for
the previous calf. The figure may be approaching half if allowance is made for calves weaned
between the mother conceiving again and being noted by us as pregnant. This continuity between
consecutive calves is apparent from field observations; many mothers are still closely associated
with an established calf (usually ≥ 2 years old by this stage) when they are suddenly seen with a
neonate. At this point, the earlier calf usually stays in the same social group as their mother but is
not so closely associated and, over time, becomes independent. If the new calf is lost, however,
the previous one usually remains associated with the mother. This continuity of care and lactation
between successive calves, and the resumption of care primacy by an earlier calf if its successor
dies, is very likely to enhance reproductive fitness in the boto, especially when neonatal mortality
is high. The same strategy is known in some marine dolphins (Whitehead & Mann, 2000).

The duration of lactation leading to the successful weaning of a calf varies by a factor of almost
four in botos. Such variation, and the range it covers (1.5– 5.8 years), is not unusual in dolphins,
however (Myrick et al., 1986; Herzing, 1997; Wells, 2014). Despite very different social systems
and environments, the duration of maternal care is remarkably similar between this freshwater
dolphin and its marine counterparts. Over a quarter of mature females were found on
examination to be neither obviously pregnant nor lactating. A small proportion of these animals
might have been in the very early stages of pregnancy, when the fetus was not visible on the
ultrasound screen, but the majority were apparently ‘resting’. Field evidence indicates that this
situation is often brought about by a young calf being lost in the recent past, but some females
simply do not produce a calf seen by us for many years. Whether this is attributable to repetitive
loss of fetuses or neonates, illness, failure to conceive or simply an unusually slow rate of
reproduction is not known. An increasing IBI with age may be expected in odontocetes (e.g. Martin
& Rothery, 1993; Wells, 2000). The failure of the calf sex ratio to stabilize with increasing sample
size was unexpected. The reason for this may be, at least in part, that this species is sexually
dimorphic in body length (Martin & da Silva, 2006) and that males are likely to be growing
marginally faster than females of the same age in the months after birth. Regardless of the reason,
it would be unsafe to infer a fetal sex ratio from these data, despite the sample size. This situation
illustrates a potential pitfall of using postnatal information to estimate this measure in any animal,
yet it seems rarely to be reported in other studies. Fiala (1980), however, showed that it is
inappropriate to use sex ratio data at any age to estimate the ratio at any other age.

Despite evolving many physical changes in adapting to a shallow, freshwater environment over
millions of years (Hamilton et al., 2001; Gutstein, Cozzuol & Pyenson, 2014) and developing
distinctive social differences (Martin et al., 2008; da Silva & Martin, 2017), the boto has very
similar reproductive characteristics to marine odontocetes. The birth of a single calf (uniparity),
length at birth, duration of gestation and lactation, the overlap between lactation and pregnancy
and the age at sexual maturity in the boto are all consistent with what would be expected in a
delphinid dolphin of the same size. In I. geoffrensis, the fundamental reproductive characteristics
of small odontocetes have been remarkably robust to profound physical and social adaptive
changes since they diverged from the Delphinids some 18 Mya (Lambert et al., 2017).